June 30, 2025 • 65 mins
The fear that AI will replace skilled professionals is misplaced – at least in lighting design. "I don't want AI to do lighting design for me," explains Juan Ferrari, Lighting Design Director at Hoare Lea, UK, "I want AI to actually make my lighting design better."
This enlightening conversation explores how creative professionals are harnessing artificial intelligence as a powerful assistant rather than viewing it as a threat. Juan, whose fascinating journey from actor to architectural lighting designer gives him unique perspective on narrative and emotion in spaces, shares practical ways his 25-person team uses AI daily to enhance their work.
From streamlining mundane tasks like email writing to creating custom GPTs for technical processes, Juan demonstrates how embracing these tools allows designers to focus on what truly matters – the creative play with light that no algorithm can replicate. He walks us through challenges like maintaining data integrity when training models, ethical considerations around copyright, and the necessity of transparency with clients about AI usage.
Perhaps most compelling is Juan's vision for the future: AI handling computational tasks to free designers from computer screens so they can physically experiment with light. "We are lighting designers. We spend quite a bit of time behind a computer doing things to enable the play, and the play seems to be a really small portion of our job," he notes. "I wish that AI enables us to do much more of that."
Whether you're a lighting professional curious about incorporating AI into your workflow or simply interested in how creative fields are adapting to technological change, this conversation offers valuable insights into maintaining human expertise while embracing powerful new tools. How might AI enhance your creative process rather than replace it?
More from the Light Talk Desk
Watch my vlogs, industry dialogues and project tours.
Listen to enriching stories, deep insights, and expert opinions in my podcast.
Learn directly from a master, absorbing the knowledge, techniques and the artistry of lighting design through my masterclass.
Sign Up to the AI Fundamentals course for Lighting Designers.
Download and delve into my works 'Light Talk, a Year in the Life of Light' and 'Light Talk 2, 16 Years in the Life of Light' - follow the journey of lighting design through his books.
Check out lighttalk.space
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
I don't want AI to do lighting design for me.
It's the opposite.
I want AI to actually make mylighting design better.
You know AI is not going toreplace lighting designers.
They're not.
Nobody will be able to bereplaced when they have that
wealth of knowledge in aparticular subject, and AI is
(00:20):
nothing more than a tool thatlearns.
So if you actually teach themthe wrong things, it will give
you the wrong result.
It's important that we trainour models in the most
appropriate manner and by thepeople that know the subjects.
Any mistakes that you make ontraining your models or your
agents then cannot be erasedfrom their memories.
It feels that you need torestart training something
(00:44):
different in order to get theright result.
All of your work can go quicklyto the bin if you actually make
a mistake in that front.
Speaker 2 (01:02):
Juan Ferrari.
Welcome to Light Talk.
Speaker 1 (01:05):
Thank you.
Thank you, Martin.
It's a pleasure to be here withyou.
Speaker 2 (01:09):
It's a pleasure to actually have conversations with
you in any platform and in thisplatform yeah it has been quite
a while since we last spoke,that's for sure, but I'm having
you here because I want to talkabout AI and the future of
lighting design.
But before we dive into that,give me a little bit of your
background.
A lot of people know you, butthere's also probably some of
(01:30):
our audience that don't know youso good.
If you gave us a bit of yourbackground in terms of where
were you born, what did youstudy, how did you get into the
position you are today?
Speaker 1 (01:42):
It's a very convoluted journey, as most of
the lighting designers in theindustry I I I'm argentinian.
I was born in buenos aires many, many, many years ago I'm not
going to tell you when was that,but many years ago, you need to
trust me there.
Um, I trained as an actor andand then I started directing
(02:02):
theater and then I actuallystarted performing and then I
started getting involved in allof the technical aspects of
theatre, and technical aspectsthat I liked the most was
lighting, and I became kind ofan accomplished theatre lighting
(02:23):
designer To the point that Istarted doing things that were
over and above what I thought Iknew.
So I came here to the UK tostudy theatre lighting design.
I studied at the Royal CentralSchool of Speech and Drama, I
did my course there, I did a BAin theatre lighting and from
(02:44):
there on, while I was atuniversity, I started shifting
my attention to architecturallighting.
I find it fascinating that thenarrative that we had in theatre
and that I have actually livedas an actor and as a director
and as a lighting designer intheatre didn't naturally exist
(03:04):
in architectural lighting.
So I actually started payingattention to that and starting
exploring that, to the pointthat I ended up working in
architectural lighting.
So I got my first job as anarchitectural lighting designer
in a company called Equationthat has now changed hands.
It's the actual equation.
It's the GIA equation, but atthat point, mark Hensman was
(03:28):
actually the director atEquation.
So I started working there.
I met some of my colleagues, mycurrent colleagues.
You had Horley there, inparticular.
Jonathan Rush and John thenmoved to Horley and I moved
shortly after, and since then wehave actually been developing
the lighting team for Hawley,first at the direction of
(03:50):
Dominic Merrick and now underour own direction.
So between me, jonathan andRuth Kelly-Wasket, we are
actually running the lightingdesign department of Hawley,
which is quite a and a very wellrecognized and well established
lighting design department inits own right within a massive,
really big engineering companythat is also part of a bigger,
(04:14):
even bigger multinationalcompany called tetratech.
So that's my role at the moment.
I direct a team of 25 people inlighting design.
They're lovely.
They're really, in my view, themost creative people that I
have actually met, with anenormous understanding of the
technical knowledge that youneed to have in order to deliver
(04:37):
the best quality lightingdesign possible.
So, yeah, that's my rolecurrently and I love talking, so
I do a lot of talks.
No, no no, you're here to talk.
I'm quite curious also, martin.
So that's why I ended upworking and exploring AI, which
(04:58):
is another big story, anotherlong story in its own right.
We'll get to that, we'll get tothat.
Speaker 2 (05:06):
I've got a cheeky question.
That's the fact that you havebeen an actor being of help in
your, in your profession, whenmeeting clients or uh, I think
that I, actually, I, I thinkthat I it helps.
Speaker 1 (05:21):
Communicating anything that you do to
understand your body, tounderstand the way that you
express yourself, to understandthe emotions that you feel while
expressing, and any sort ofregister that you get of your
own tools, of your own physical,emotional tools, is quite
helpful in any environment.
So, yes, it helped me it.
(05:43):
It, to be honest, um, one ofthe most important things for me
is to to be able to communicateproperly, and lighting is a
very difficult.
It looks very easy, but it'svery difficult to communicate in
words, and and I think that my,my theater background have
helped me enormously, and thenthe notion that lighting is a
(06:07):
tool that actually tells storiesand that has a narrative.
All of that piece of work ofmine is based on my experience
in theatre really, so yeah ithas helped me a lot, yes.
Speaker 2 (06:20):
Yeah, I think it's a great combination because in
theatre lighting your focus istechnically on the actors when
you light, but in architecturallighting it's more about
lighting the architecture.
So I think that combinedknowledge I think is great.
Speaker 1 (06:37):
Also in architecture, it's not only lighting the
architecture, it's lighting forpeople using it.
So, although we don'tnecessarily experience our lives
as a play or as a narrative, wedon't feel that our lives are a
story.
We are actually living throughemotions, in the same way that
(06:59):
we are living through emotionsat play.
So it's quite important thatlighting takes into account the
feelings and the journeys thatpeople have within the
architectural spaces, and that'sone of the most important
things for me.
Speaker 2 (07:12):
So let's jump into AI straight away.
At what point of time did youget caught by the AI bug?
Speaker 1 (07:21):
I actually it's quite funny I do a talk about AI and
I actually expose myself quitein a big manner in that
particular talk because I'm auser of AI and user and abuser
of AI.
So I started playing with AI ona journey to Birmingham
actually were we were in the carand and particularly with chat
(07:45):
gpt um.
We were in a car a few yearsago and one of my colleagues
that was traveling with meshared with me this new thing
that have come out called chat,gpt and um, and I asked chad gpt
a particular question about uhlighting and he'd answered it in
a way that I couldn't answer it, even if I was given an hour or
(08:10):
two to respond to thatparticular question.
I went like, wow, this is quiteimpressive because in a
microsecond, all of a sudden Iget an answer that was quite
solid and it was very muchtechnically um and I thought
this is quite interesting.
And so the iald was doing theirenlightening conference and
(08:32):
calling for papers and Iactually composed a paper
completely with ai about ai andlighting, but that without me
knowing a lot about ai, and thepaper got selected.
So all of a sudden I had tolearn about AI in order to
present about AI, which is awhich is a an unfortunate,
fortunate situation.
(08:53):
You know, it was a little bitcheeky from me and that was
quite successful talk because,as I say, I exposed myself into
a point in which I said, look, Idon't know anything about AI, I
just put this into AI and Iprepared a presentation all
through AI.
And then this is my story Ifyou fast forward, this was three
(09:13):
years ago now, almost two and ahalf years ago, and if you fast
forward to today, I am actuallyconstantly training myself on
these new tools and in in aithat are available to us.
I'm playing with all of them, Iuse them on a regular basis.
I I encourage people.
(09:36):
I think that my mission on this, on this particular front, is
to encourage people to use it.
Some people are quite scared.
They have ethical concerns,they have moral concerns, they
have all sorts of concerns, andthe only way of actually making
this tool a less uh, a lesserproblem is using it, you know.
Speaker 2 (09:55):
So I'm encouraging people to use it I'm a bit like
you because I mean, I just gotlate last year something I I got
in touch with it and then, um,I I know if you have seen, but
I'm I'm promoting an ai coursefor for lighting designers,
which I'm doing with an aispecialist which, like you, he's
(10:15):
an architect, he's been in thisfor the last couple of years,
he knows everything of ai, hedreams and and lives ai.
But I I feel also a bit bitreluctant and a bit overwhelmed
by all the possibilities.
And you talked about the toolstwo years ago.
Well, you see what speed theyevolve and develop, so what's
(10:35):
possible today?
But for me, the reason to jumpin is to be the reason to be.
You see, I've got 45 years ofexperience in lighting design,
so I can actually sort of beingthe challenger and the reasoning
behind what AI throws at us andlook at it through the lighting
design expertise that I haveand say, well, is this correct?
(10:57):
Yes, and, like you say, whatcomes back is sometimes amazing,
like really the knowledge.
So, yeah, I think it'simportant to have that balance
and you can't better learn, Ithink, than really diving into
it and embracing it, becausethere's no way we can go around
it.
It's there.
(11:17):
It's there for us too.
It will be like the mobilephone.
It's something that's going tobe there as a new revolution,
and if you're not in it, you'renot going to win it, that's for
sure.
Speaker 1 (11:28):
So talk about tools.
Yeah go ahead.
Yeah, I think that is portant.
I think that you need to beable to use it and to be in it.
There is somethinggenerationally.
It's a really interestingconversation, the generation
conversation.
There is somethinggenerationally.
It's a really interestingconversation, the generation
conversation.
It's like we did within ourlifetimes.
We went from a pencil and apiece of paper to a typewriter I
(11:56):
would put myself in thatcategory To a computer, to the
internet and to AI, and we didthat.
I'm going to disclose my agenow I'm 50.
So within 50 years, I've donethat.
I'm going to disclose my agenow I'm 50.
Yeah, so within 50 years, I'vedone that.
My journey took 50 years.
If you think about it.
Our kids nowadays learned allof that in a period of 10 years.
(12:19):
So imagine how much all of thistool-related relationship that
we have with our profession anyof our professions will evolve
in the next 40 years for them.
What will come, we don't know,but it's going to be definitely
at the pace that we arecurrently exploring it.
So it's going to be magical,you know.
So it's incredible the amountof change that is ahead of us.
(12:45):
So what we need to build inourselves is the possibility of
being flexible and grab thesetools that are given to us.
And today today, yeah, we're in2025.
The tool that we are using, or alot of people are using, is
Chak, gpt or DeepSeq, buttomorrow it will be another one,
and but tomorrow it will beanother one, and the day after
(13:07):
it will be another one.
So what you need to be able isto become permeable to all of
these tools that are coming toyou, to be able to use it, and
the way of really using them ishaving a reason to use them.
So I think that the second bitof an NEI conversation is what
do we want to do with them?
(13:28):
What do we want to do with AI?
Because if you don't have ananswer of what do we want to do
with it, then we cannot use thattool.
Speaker 2 (13:36):
Okay, so tell me, what are we going to do with AI?
Speaker 1 (13:40):
What are we going to do with AI?
What are you doing?
Maybe?
Speaker 2 (13:43):
better.
What are you doing at themoment with AI and what do you
think we should probably bedoing?
Speaker 1 (13:50):
So let's start with a very basic point of what do we
do with AI?
The first thing that we do withAI is try to streamline our
mundane tasks.
You know, those things that wedon't like doing, like writing
emails.
What do I use AI for?
To write my emails.
I turn not to write any moreemails.
I actually tend most of thetime to dictate an email to an
(14:13):
AI.
The AI composes.
I review that email send.
Speaker 2 (14:19):
I still need to correct, you still have to
prompt, so that means you stillneed to write something.
Speaker 1 (14:23):
No, I do that verbally.
It flies into the prompting.
So I do that verbally, I do iton the microphone, so I actually
dictate it to okay all right,okay, okay, yeah, I don't, I
don't, I don't write it.
I actually go um, write an emailto martin class and I want to
actually discuss this nextinterview that we're doing
together.
I want to know the subject thatwe're going to discuss.
(14:44):
Send and the AI does it for me,formats it for me.
When you're working on yoursecond language and English is
my second language.
This is a really, reallytime-saving tool.
It does save me loads of timeand I absolutely adore the fact
(15:04):
of doing that mundane task forme.
I use it quite a bit forresearch, so I build up project
folders within within the AI, myAI of choice, and actually then
say well, look this, thisparticular bot is going to act
as a lighting agent or alighting person, and and then it
has this amount of experienceand actually based all of his
(15:26):
knowledge on this, this, this,this piece of information.
And then I ask say that I wantto research I don't know um the
the of anything, what, what didthe dinosaurs do before they die
?
Basically, I asked the AI totrain as a dinosaur expert.
(15:50):
Then it would tell me back, itwould actually start giving me
facts of what were the lastthings, what was the last supper
from the dinosaurs, and all ofthat, doing research, doing you
do that with.
Speaker 2 (16:04):
NGPT research, or are you using other tools?
Speaker 1 (16:07):
I I use I currently use because of my, of my company
policies.
We are quite restricted of whatwe can actually use or not use
for work.
So I obviously experiment on apersonal computer, but I also,
when I'm working, I have a chatgpt enterprise which is probably
the latest version of chat gptwith a full package around it.
(16:31):
So it has dali, it has saw, ithas other other cameras and
other and other um and otherservices that chat gpt brings.
But it's actually ring fencedso the information is actually
kept confidential.
So I can actually handleconfidential information within
ChatGPT Enterprise withouthaving to share it with the
(16:53):
world, because obviously, as youwould understand, we do things
that are confidential in ourjobs and our clients would be
happy if we disclosed thatinformation into different AI
services.
So I use that quite a lot.
I experiment a lot with thevisual side of AI.
So basically, I use Midjourney,I use Runway, I try to explore,
(17:18):
I'm trying to get to a point inwhich our lives become our
presentation life.
You know that we we have donemany presentations and we we
actually love the movement inthe presentation and we love the
visual engagement in thepresentation, and we live in a
world of instagram now.
So we live in a world thatimages move and flick and move
(17:41):
quite quickly and and I look atour lighting presentations
generally and they're quitestatic.
So I'm trying to get a littlebit more of a different way of
presenting through AI at themoment, which I'm exploring and
I have to admit, martin, I havenot refined this, but I have
actually got to the point that Iactually don't present static
(18:02):
images anymore.
I kind of animate them throughrunway in my presentation.
So I use that quite a bit andit's quite interesting.
The problem with images andmoving images and generating
moving images in AI is that itstarts deforming objects and I
know there are some certainpeople that do it in a marvelous
(18:23):
way.
I'm not one of them, but I'mgetting there.
I'm not a.
I know there are some certainpeople that do it in a marvelous
way.
I'm not one of them, but I'mgetting there.
I'm actually playing around andtrying to get to that sort of
image movement within the AIworld that I'm exploring.
And then there is the use for AI, of which would be probably the
(18:44):
iterative design conversation,when you have to bring a lot of
options, so giving the machinean option and say, look, give me
the lighting design for thisroom.
Or I assume that I want alighting design for this room
with downlights and offer detail, and I want some up lighters
here and there.
Just give me some other optionsand it would quickly give me
(19:06):
options on the way that Iactually could approach the
lighting to that space.
Now I could think about thoseoptions on my own.
Obviously, I've been doinglighting design for so many
years, but sometimes the thequirky nature of ai, the
hallucinations of ai, actuallyopen another dimension, you know
, and give you yes, give you anoption that you have not thought
(19:26):
to and it's kind of ridiculous,but it's lovely at the same
time, you know.
So it's.
It's that, it's another thingthat I use?
Speaker 2 (19:32):
do you challenge your designs as well using ai?
Because I think that's anotheropportunity that we now have is
just feed the design into the aitool and ask it to challenge.
We are currently doing researchand I'm talking about the AI
course that I'm developing atthe moment.
One of the things that we aretrying to figure out is whether
we can input, like a dialectscalculation for argument's sake
(19:59):
and then see whether AI can saycan we improve this, how can we
make this better.
See whether AI can say can weimprove this, how can we make
this better.
But I think it would hold forany sort of design proposal that
you make that you can actuallychallenge your own design
creation.
It's like how could we improvethat or are there any weaknesses
in this?
What works or doesn't work?
(20:20):
Are you doing that as well?
Speaker 1 (20:22):
Yeah, we do that.
So we do that in artificiallighting and in daylighting too.
So we use the basics of what wehave been doing pre-AI in
parametric design with Rhino andGrasshopper in order to, for
example, just actually build upa room and say what's the most
efficient opening that I canhave as a window in terms of, I
(20:43):
don't know, thermal comfort andand um, sorry, um heat gain and
and um and uh and uh, daylight,um, daylight, um, penetration
and and basically we have used,I don't know, grasshopper and
and rhino for many years in ingiving us those numbers, you
(21:04):
have to think that we are anengineering firm, so our numbers
are quite important to us inmany ways and so we have been
playing with them for many years.
So AI actually brings anotherlevel of joy to that and another
level of speed.
And, yes, definitely on theartificial lighting side, if you
don't remember wrong, reluxused to have this tool that you
(21:26):
can give them in a, in arectangular box, not the most
complex of geometries, but in arectangular box.
It would give you the the mostefficient way of actually
placing your luminaires within,within, within your plan, and
and I think that that was pre-ai, you know.
So, yes, we, we have beenplaying around with that.
A challenging design through aiis quite important, in the same
(21:47):
way that doctors challengetheir findings with AI.
You know, you have heard thatdoctors, for example, they have
a scan and they're actuallysearching for I don't know
tumors or things like that, andthey go like, well, have I
actually looked at this x-rayproperly?
And the AI will actually say,well, you haven't looked at this
bit here.
Actually you might want to lookat that, because that's kind of
(22:10):
not very common.
But in order to actually get tothat point, martin, as you would
know, you have to train yourAIs, and that's a critical thing
.
An AI is nothing more than atool that learns, so if you
actually teach them the wrongthings, it will give you the
wrong results, in the same waythat if you input the wrong
(22:30):
parameters into a lightingcalculation, you will get the
wrong results and it will bemeaningless.
So it's important that we trainour models in the most
appropriate manner and by thepeople that know the subject.
So any mistakes?
What I do find out is that anymistakes that you make on
training your models yeah, oryour agents then cannot be
(22:54):
erased from their memories andit keeps on repeating and it
feels that you need to restarttraining something different in
order to get the right results,so all of your work can go
quickly to the bin if youactually make a mistake.
On that front, I think thatdata sets are quite critical, so
(23:17):
basically, having the rightdata to inform these AI agents
is absolutely a must.
Speaker 2 (23:25):
You mentioned training, so are you the only
one in the company doing that?
I would assume that you're notthe only one and that you have
your own team on this.
I can also imagine that you Idon't know whether you have like
an office model and a personalmodel, because you can train, I,
on, on, on on your personal uhdata that you want, uh, whatever
(23:49):
I I tool to follow.
But if you use an enterpriseversion, I don't have an
enterprise version, I have justthe pro version.
But yeah, um, you, you train, I, I'm trying, I'm just, I'm just
a starter, so I'm I'm trainingit all myself.
I can imagine that if you havean I don't know whether the
enterprise works as a collectiveso that multiple people can use
(24:13):
the same tool, so the trainingis more.
Speaker 1 (24:17):
You can create a bot and then share it with other
people.
So, yeah, you can actuallytrain an agent or a bot and then
that share it with other peoplewithin enterprise.
You can do that.
You can do that in pro.
I can actually shareinformation with you that I want
to share with you, or a botthat I want to share with you.
That that shouldn't be aproblem.
Yeah, they are compatible so inthat respect.
Speaker 2 (24:40):
So I can actually do things from pro to pro, can you
maybe explain a few of thetraining things that you're
doing?
How do you train that to doexactly what you want it to do,
or to make it machine learn whatyou want the AI tool to become?
You talk about agent.
(25:01):
Are you really already creatingan agent or is it just a way?
I mean, people talk about agentas an autonomous tool, but I
would imagine in this case,you're talking about an agent as
something that you have trained.
Speaker 1 (25:17):
Yeah, more starting with a bot concept and then
trying to get it to go to anagent.
But the bot concept isbasically you have your GPTs.
Yeah, just create your own GPT.
It's exactly that.
That's what I'm trying to say.
What we do is we create our ownGPT.
A GPT, for example, can be, saythat you wanted to create room
(25:39):
data sheets, sheets.
So what you need to embed intothe bot, into the GPT, is the
knowledge of where they need toactually grab the parameters in
order to create a room datasheet.
So you will actually connectthe knowledge of the AI to the
(26:03):
standards that you need toactually comply with.
And you will feed all of thatand you'll say, look, on the
room data sheet level, I need toget this parameter.
So, basically, I don't knowilluminance levels, glare, yeah,
all of the parameters that weuse in lighting.
And from there on, the machinewill actually start
(26:27):
understanding how to get you tothat point.
So, basically, then you willupload a drawing.
It will say you have I don'tknow you have.
Say that you're in a hospital,you have a clean room, you have
a treatment room, you have a Idon't know an operation theatre.
Yeah, then it would justtabulate in a table very quickly
(26:49):
all of the levels and all ofthe parameters that you need to
hit within each one of thoserooms.
That's an enormous time-savingbot, for example, for us.
So basically we use that on aregular basis.
Speaker 2 (27:04):
But I can imagine you can also create your spec
sheets and other documentationas templates that you can just
call back if you want to.
Speaker 1 (27:19):
We are quite so.
We have actually done aswimline diagram of all of our
processes.
So this is outside of AI.
So we've done lighting design.
What do we do From the momentsomebody tells us we have a job
for you?
Speaker 2 (27:39):
We would like to discuss this.
Speaker 1 (27:41):
What do we do?
What sort of information?
What sort of information goesout in what sort of program?
Who gets involved in it?
So we have actually done aswimline diagram of all of our
processes and we are trying,against each one of those
processes, different AI tools.
Now we haven't yet developed atool, an AI tool that assists us
(28:04):
with each one of thoseprocesses.
We've been miles away from someof them, which are the ones
that we would really like tohave.
So, for example, I would verymuch like AI to assist us with
the Revit integration of ourdesign.
So if we had a model, if,ideally, we have a model from
day one how AI can actuallystart populating things in a
(28:29):
logical manner for us, becausethat's a really time consuming
exercise for anybody at themoment and we're trying that.
That relationship between thecalculation program and Revit
and how they link together.
We seem to be modeling andmodeling, and modeling and
modeling, and AI doesn't seem tobe able to help us.
So we do a model for acalculation and we populate a
(28:51):
model like if it was a drawing,which is really uncomfortable.
And why are we not using thesame model?
So that streamlining all ofthose processes and getting AI
to help us with all of thoseprocesses would be the future.
That's what we're exploring.
But in essence, on thatstreamline diagram that I was
talking to you about, with allof our processes, what we are
(29:12):
testing is against each one ofthose processes and what can we
use and what technology can weuse and what tools can we use in
order to actually make themmore efficient and better?
Speaker 2 (29:23):
And I think the thing is that today it might not yet
be possible, but tomorrow itwill be.
I mean, it's developing andevolving so fast that, um, it's
whatever we have in mind that wethink will be able uh, in in in
the near future will happen.
I'm pretty sure of that.
Um, it's just a matter, likeyou say, you need, you need to
(29:43):
get your hands dirty and andplay with it and explore, but
then you have also, I mean,exploring things within your
enterprise platform is one thing, but then exporting it to a
client is another thing.
So I would imagine that youhave some checks and balances
(30:03):
there in how, what is it thatyou can use towards your client?
In what is it that you can usetowards your client?
You probably need to excerptthings and maybe some you don't
want the client to have and somethings you might think are
private to you.
Speaker 1 (30:20):
So how is that, martin?
In that respect, we're quitetransparent with our clients.
We have systems in place tomake sure that all of our
information, in the same waythat you have IT systems, make
sure that all of our informationdoesn't get leaked and doesn't
get compromised, and thereforethat's why we work with
(30:43):
enterprise, which maybe is alesser it's not the pro version,
it's probably a step behind butwhat gives us is a ring fence
of the information and makingsure that everything is secured
within our own service.
So that is quite critical forus.
Speaker 2 (30:58):
So and then you know what I meant I meant to upon is
that you, if you do it on anindividual basis, you control
what you put out there, but ifyou have a team, you still need
to have a unified company uhrestricted or company managed uh
output and and the bigger yourteam, the more complex that
(31:21):
might be, because people may goin different directions.
So I mean standardizing the wayyou present to your client.
That was more what I meant.
Speaker 1 (31:30):
Yeah, well, it's part of the quality management that
we actually do.
So any company, any largecompany, you would know.
We have quality managementrules and basic quality
management procedures that weneed to follow, and we actually
do the same with AI.
So AI is no different to anyother things that we do.
We had to.
(31:53):
There is something it's a bitmore complicated to control In a
world that is never evolvingand you want people to explore
tools.
It's very difficult to do itwithin a multinational company.
So that's where you actuallyhave dedicated exploration and
research machines and then youhave your day-to-day machine in
(32:16):
which you can do your job in asecure manner.
But that comes onto theprofessional.
The way that we handleinformation needs to be very,
very much controlled, because ifnot, we would be making things
that we don't want to leak.
So, uh, but, but.
But at the same time, you wantto encourage your design team
and your people to explorethings.
What?
(32:36):
What actually surprised me andI'm going to go back quickly to
one of your questions I don'tthink I answered which is you
said to me did you, does?
Your whole team works with aiand we work at very different
levels, you know, and in verydifferent ways and I encourage
them to work with AI.
I actually go like please usethem.
We have loads of options withinthe suites that we have within
(32:59):
Holi for them to actually playwith, but sometimes they don't
find it's like with everything.
Sometimes a tool is not usefulto them and they actually might
choose to do it in a differentway and without the assistance
of an ai.
So that's, that's a choice, youknow, and and um, I don't think
you can push people into ai.
I think it's a pity if people donot go into the eye because,
(33:22):
exactly, martin, it's the samething that happened with I don't
know autocad.
You would actually said topeople at some point there were
people drafting drawings anddoing lines and working on it,
and, and, and those peoplethought that they could keep on
drawing by hand and some of themwere reluctant to actually
(33:44):
engage with AutoCAD and at somepoint nobody was drawing anymore
and everybody was drawing inCAD.
So it happens with every singletool that comes into the market
.
People would have said the samething from the internet, from
Photoshop, from everything, thecalculation programs.
Speaker 2 (34:00):
People used to do lighting calculations by hand.
Yeah, some people are morecomfortable than others.
I mean some are sort ofnaturally flowing into it.
Others I mean I'm literally oldgeneration, I started when we
didn't have are more comfortablethan others.
I mean some are sort ofnaturally flowing into it.
Others, I mean I'm literallyold generation, I started when
we even didn't have computers.
So you know this my first, myfirst encounter with the
computer was a huge mainframethat was sitting in a big room
(34:20):
where you had to do punch cardsand things like that for
lighting calculations.
But that's well, okay, we talkI'm.
I'm slightly older than you.
I won't disclose my age, but 51.
So how many people would yousay are working on AI now,
percentage-wise?
Because of course I imagine youstill have a bit of a dedicated
(34:44):
team.
Yes, you can encourageeverybody, maybe even on email
level, but even actual designproduction, that might be a
different sort of approach.
Speaker 1 (34:57):
How many people from my team, or how many people
within Holi, or how many peoplein the world?
Speaker 2 (35:03):
Okay, well, what is within your?
That's a really interestingquestion.
I actually Just expand.
Well, what is within your Of myteam?
That's a really interestingquestion.
Speaker 1 (35:11):
I actually I don't have a.
I have to guess the answer ofall of the three questions.
Obviously, the world one is Idon't know it would be of the 7
billion people that we are.
I don't know how many peopleyou say I guess that a lot, but
on my team.
So I have 25 strong in our teamand I think on a daily basis, I
(35:36):
would say that probably 70% ofour team engages with AI.
Pretty sure that everybody hasengaged with AI in one way or
another within our team and somepeople do more regularly than
others.
As I say, some of us use it allthe time.
Speaker 2 (35:56):
Yeah, I mean it's a bit addictive, let's face it.
Once you get onto it, it's likeit's hard not to use it.
Speaker 1 (36:05):
Absolutely, absolutely.
I use it as a search engine.
I don't search anything else inGoogle anymore, I just search
in ChatGPT.
I find it easier.
I find that the questionseasier, the answers easier.
Speaker 2 (36:20):
You know, even if you use Google search, you get into
Gemini, it comes up with.
The first thing that comes upis a Gemini answer.
Right yeah?
Speaker 1 (36:28):
I don't even like the expression of Google anymore.
Basically, google gives me Idon't know 100 results with
three or four ads in the middle,and I've got to go.
I need a phone number and I goto Chachipiti.
I dictate to Chachipiti I needMartin Klaassen's phone number
and it will just give me yourphone number.
Not a lot of ads and a lot ofyou know, soon you can even use
(36:50):
my voice.
Speaker 2 (36:53):
I'm currently cloning my voice.
I have the first clone and Iwant to use it to do an e-book,
so that you know my books.
I can read them, but I don'twant to read them.
So I'll get my cloned voice toread my books, that's very good.
Speaker 1 (37:08):
Writing a book is one thing, you know.
I find that fascinating.
What you're telling me is soenriching, Martin, because it's
endless the amount ofopportunity.
I've never thought about it.
Actually, I particularly have aproblem with reading, because I
read very badly.
Imagine when you're reading ona second language.
My eyes go completely crossed.
(37:30):
But I adore the narrative of abook, so I adore the stories
behind the book.
Now, what you just told meopened my eyes, like that, you
know, because it's great and Inever thought about oh, why
don't I actually train my AI tohave my voice and then give them
the books and read them back tome?
That's perfect.
It's an audio book with my ownvoice, which is great.
(37:51):
Yeah, yeah, yeah.
Speaker 2 (37:53):
Well, soon you can read my book or listen to my
book with my voice, exactly,exactly, remind me, I'm like you
.
I followed an AI course earlierthis year that really opened my
eyes.
It was supposed to be athree-hour course.
It ran out to nearly four hoursand the opportunities and the
(38:18):
things that were possible, thatblew my mind away.
And that's when I dived head into develop this AI course,
because I think people need tounderstand what's possible and
get over the threshold of angstand the threshold of fear of
what it all is and just see itfor what it is.
And then, of course, you needto make up your mind of what's
(38:41):
good and bad about it, and notforgetting that it is an
assistant and you really need tomake sure that you keep the
creative control and thecreative mastership about what
it all creates Right.
But yeah, I think theopportunities are endless.
Speaker 1 (39:01):
So I am doing a course called AI for Business
Value and it's actually a mixbetween AI and a business course
, which is quite interesting.
It's on the right, it's anapprenticeship by Multiverse,
which is a very basic course onboth business and AI.
But one of the things that isquite good about that course is,
(39:24):
first, that everything in thecourse, within reasons and
within the right parameters andthe right way of doing it,
should be and could be run by ai.
You know so.
There is no.
There is no.
You need to read this book.
It's you.
Do whatever you want.
You.
Give me the solution howeveryou want.
You, use the hour however youwant you, even if you want to
play back everything that aisays to you.
(39:45):
Obviously, there is a risk indoing that and we all know it.
That's completely acceptable.
So there is no.
This is a task that you shouldbe doing on your own without AI.
Ai is there.
It's like if nobody's notallowing you to use a pen, you
know, here's a pen or a pencil.
Ai is a pencil.
Use it.
And one other thing that welearned is that notion of making
(40:10):
sure that the data sets arecorrect and that you're the
reasons why you're using AI,that you have a reason to use it
.
What do you want to do?
What bit of the lighting designprocess do you want to
streamline and then find theright tools to do that.
Speaker 2 (40:29):
It's quite an interesting course well, I I
started, I'm still lovesketching and I, I I'm on the I
don't know if you have the samething, the remarkable tablet
where I can still draw, but it's, it's now digital, right.
Yeah, you know the mark.
So, and now I've started to doa sketch and then input it into
(40:50):
an AI tool and ask it to developit further, just from my sketch
.
Yeah, so there's also endlessopportunities.
Speaker 1 (40:58):
It's great, it's absolutely great.
So, for example, one of thethings that you know when you go
into a meeting, you take notes.
Yeah, yeah, and sometimes youdo take notes by hand and
sometimes you type them in.
I tend to take notes by handwhen I'm in a meeting, but
(41:18):
before I actually used to lookat my notes and type them back
and play them back to I don'tknow a client or my colleague or
whatever.
Yeah, I now take a picture ofmy notes, upload it into chat
gpt in this case and I extractthe text and I have them all
typed in.
Speaker 2 (41:33):
Yeah, Well, you can also ask to make an audio
version of it so that people canlisten to it.
They don't even have to.
Yeah.
Speaker 1 (41:41):
I haven't done that but, yes, we can yeah.
Speaker 2 (41:43):
Yeah, yeah, yeah, those are all of them.
I interviewed someone recentlyand it was also about an hour
interview, but I asked NotebookLM to do a short version of it,
like less than 50 minutessummary, and then you get two
people talking to each other and, in a very, very concise and
(42:05):
great way, you can just listento the summary.
In that way, you don't evenhave to look back.
I wanted to ask you somethingabout copyright and intellectual
property, because obviously, asdesigners and creators, we
always look at that element.
How do you see that within theworld of AI?
Speaker 1 (42:27):
It's quite a complex subject, isn't it?
Because AI in Europe, there'sthe EU directives on AI at the
moment, so there is, there ismeaningful legislation, but
there isn't, there isn't in theUK at the moment.
So it that it's, it's.
We're evolving on all of thatsubject, you know, because I
(42:49):
think it's quite critical thatthere is a degree of ownership,
especially when it comes tocreative material.
I think there is something inthere that needs to be
legislated and I think it willbe legislated properly.
If you ask me, I'm quite carefulwith that, because when you're
(43:11):
extracting and you're generating, I try to generate my own work.
You know, I've actually tried,but I'm sure that the model have
learned from other people and II don't.
I wouldn't be able to tell ifI'm actually infringing a
copyright or not on a particulartext, because I wouldn't know
what the combination of wordsthat other people have used to
(43:31):
actually talk about a particularsubject.
So it very much everything thatgoes out and I stress this out
everything that goes out thatwas created by AI needs to be
stated, as with AI assistance.
You know you need to.
Actually we need to write intoour, into a feedback.
(43:51):
Yeah, you need to declare it.
I know you need to.
Actually we need to write intoour, into a fee.
Yeah, you need to declare, youneed to declare, yeah into into
the, into the contracts that wego into, that we are going to
use ai in that work.
Speaker 2 (44:04):
It's absolutely fundamental that that is the
case so your case is actually inyour, in your contract.
You already put in yourcontract.
Speaker 1 (44:12):
We state that we would be or we could potentially
be using.
Yes, we just need to startdoing that because, if not, it
would be not being very clear onthe way that we are working.
No, I think we need to behonest and transparent about
these things.
Speaker 2 (44:29):
Yeah, I agree, and now is the time to start
mentioning that, I think,because most people are starting
to get familiar with it and youknow there's also ways to
potentially protect your work incase of non-payment, like now.
We could potentially I don'tknow whether you, but it is
(44:50):
something that's being exploredby putting all your deliverables
in the cloud with asubscription link and if you
don't pay your subscription, youdon't have access.
You cannot download it, you canonly work in there.
You can see everything, but ifa client, for argument's sake,
would not pay, then you can'taccess the work anymore.
(45:13):
We all experience situationswhere clients are tough
paymasters and sometimes theywant everything for free, or
they want everything and theydon't want to pay, or they are
very difficult in payments.
But potentially, through AI andto this kind of digital and to
this kind of digital cloudsituations, we could better
(45:34):
manage that our whole paymentschedules and all that.
Speaker 1 (45:41):
Yeah, I suppose that's probably something that
other companies would startactually exploring.
I wouldn't be surprised ifcompanies that have to do with
legal, with legal, I don't knowlaw companies, law firms are
(46:03):
actually already looking intohow that works and how they can
bring a service on the back ofthat.
Yeah, yes, in terms ofcopyright, just to wrap up the
idea, it's very simple, Martin.
Everything that I send is myresponsibility, regardless of
how I actually create it or not.
You know, whatever I used tocreate it, if I press send on an
(46:25):
email, nobody's pressing sendfor me on the email.
Yeah, so if I created an emailwith AI that says all sorts of
horrible things.
It's me sending it.
So the one thing that we needto be careful in the same way
that we are careful withanything that we do is that when
we actually press send, that isa piece of information that we
agree with.
(46:45):
There is no point of blaming.
There's enough law casesalready.
Speaker 2 (46:52):
It's part of your already.
It's part of yourprofessionalism, part of your
integrity, you know, andtransparency.
I think in order to do business, you need to be in integrity,
also for the longevity of yourprofession and your
relationships with the client.
I think, yeah, and also it'svery dangerous.
Speaker 1 (47:12):
It has.
There are many law casesalready of people that actually
say no, well, sorry, I wasactually doing this with an ai.
I was well, who cares?
Speaker 2 (47:18):
you know, it's your, it's your responsibility, uh,
yeah, yeah, exactly, it's likesay look, I didn't draft this,
because I was drafting it in inin autocad or in revit, or I was
actually creating this model inrevit and I'm sorry, it's not
it's not real because it's yourresponsibility Exactly, and it
takes years and years to buildup trust and respect, but it can
(47:38):
be destroyed in minutes, so youreally have to be very careful
on how you treat yourrelationships in that respect.
I want to make a littlesidestep to Lux Futurum.
You are a guest speaker on theLux Futurum.
You are a guest speaker on theLux Futurum platform.
You will be doing one of thepromo tours in China, and thank
(48:01):
you for that.
What does I mean?
You know a little bit more aboutLux Futurum now.
It's a platform and arecognition program that aims to
recognize future ideas, futureconcepts, future projects,
products, things that sort ofembody the spirit of the future,
new innovations, and we reallywant to bring that foreground
(48:25):
and give people the ability andopportunity to promote and share
their ideas.
And not only that the winnersthat are being shortlisted are
then invited to the final eventat the end of the year to
actually present, because wefeel that the learning
experience and the knowledgesharing about why something was
(48:48):
a selected winner is asimportant as just selecting it,
because we want to share thatinformation.
What is it that made that sospecial?
Um, I just want to hear a few,a few insights from from your,
from your point of view, uh,what you think about luxury
juror and and why you're so uhnice for us to support it oh,
(49:11):
first I want to actually thankyou for the invitation.
Speaker 1 (49:15):
I think supporting any platforms that are actually
looking at innovation is key toany development of any
profession, not only in lighting.
I also really believe in thecrossbreed of specialisms in
terms of how they integrate, sothis sort of innovation
(49:38):
platforms will allow people tostart thinking a little bit
outside the box, and not only inlighting terms, but in lighting
terms connected with otherelements and how they all, and I
think that that brings anotherlevel of innovation and design
and development.
(49:58):
I've heard some reallyinteresting talks about people
that were doing innovation innot only lighting design terms,
but in artificial intelligenceterms in the 1970s, intelligence
terms in the 1970s, and ifthose people would have had
these platforms at thatparticular point, maybe we would
(50:24):
have actually engaged with AImuch earlier and we wouldn't
have the winters of AI, as theycall them, where people just say
, well, it can't be bothered byAI anymore and they'd stop
researching and they'd stopdeveloping because we lost 20
years in doing that, you know.
And now, 20 years losing 20years or not losing 20 years,
who cares?
But I thought that we wouldprobably have a different way of
(50:46):
working already if AI wouldhave been developed around the
1970s and 80s, instead ofactually being stopped and
frozen on an AI winter.
You know, very likely thoseplatforms, these platforms like
Futura, are the places forpeople to speak, to amplify what
(51:07):
they're doing, to capture moreimaginations, to capture more
funding.
To capture more imaginations,to capture more funding, to
capture more track speed, and Ithink that's why I support them,
you know, and that's why I liketo participate in them.
They enrich me, martin, that'sthe thing, you know.
I listen to people innovatingspeak and they enrich me my
(51:31):
persona.
I love that.
I actually have to say thatit's great because, in the same
way that this conversation hasenriched me with the, you know,
when we're talking about yourbook, that you're reading your
book to yourself.
It's great, you know it'sfantastic.
It is fantastic.
Speaker 2 (51:48):
You'll be cloning your voice very soon.
Speaker 1 (51:51):
Definitely.
Why not?
Why not it's?
used in some medical you knowthere are some really bad health
conditions that actually peoplelose their voices and the
people record their voices andbe able to play them back.
How good is that, you know how?
How really good is that andactually to be using it without
(52:14):
my need.
Is not that because I don'thave a voice?
It's because I really don'tlike the process of reading.
I love reading, but I don't.
I find it painful in my eyesand in my head.
Yeah, yeah, yeah, and it'sgreat that somebody, or my own
voice, reads it back to me, youknow, or my own voice reads it
back to me.
Speaker 2 (52:32):
You know, interestingly, in last year's
submissions for Lux Futurumthere was hardly any AI-related
submission Quite surprising, butI would suspect that this year,
with AI taking such a flight,that we would have more
AI-related entries.
I hope so.
(52:53):
I haven't seen any yet with AI,but I would imagine and this
year we're also specificallyasking all the entrants to list
and declare AI tools that theymay have used in the submission,
because I think, for clarityand also for educational
purposes, it's important tounderstand what is humanly
(53:15):
created and what is created withthe help of AI tools.
Yeah.
Speaker 1 (53:24):
I think it's actually good to actually I think it's
necessary to declare what you'reusing as a tool.
The reality is that I wouldguess that most of the people
have used AI last year.
They just didn't declare it or,in some cases, they even hide
it because they think there is anotion and this is a really
(53:48):
interesting bit of aconversation which is a notion
that everybody that is using AIis cheating.
We need to move away from that.
Everybody that is using AI isusing a tool that is powering
and amplifying what they'redoing.
They're not cheating, you know.
Speaker 2 (54:08):
I think, if you declare that sort of idea of
cheating, because you know I'mhonest, I'm using this.
I'm still the creator-in-chief,but I'm using this tool to help
me.
Another point that I wanted tohighlight is that one of the
goals of Lux Futurum is also tobridge East and West.
You know China is thebirthplace of.
(54:30):
You know where mostmanufacturing of the lighting
technology is being done.
Most of our light is all comingfrom china.
90 is all made there and, ifnot, the full feature, the
components are coming from there.
Um, and there is notion also,like you said, that oh, china,
they are copiers and they're notreally, um, you know, at the
(54:51):
forefront of everything.
But that has changeddramatically.
China is very much aninnovative country in that term.
They, in some aspects, areahead of the Western world in
terms of digital integration,use of AI in many applications,
(55:13):
and we felt that there's thisapprehension about China, for, I
mean, I've been coming to Chinafor many, many years, so I
don't have that.
I know these lovely people andthey have moved things forward
quite dramatically, and I thinkthe ability to connect East with
West through this platform isalso what drove us to develop
(55:36):
this.
I would imagine that that'ssomething that you can see.
Speaker 1 (55:42):
Learning from others, from other cultures, from other
people, is absolutely criticalfor any big shift and big
development.
I always use the same exampleChristopher Wren, which is the
architect of St Paul's Cathedraland the Sheldonian Theatre in
(56:03):
Oxford.
One of them did the master planfor London after the Great Fire
.
He is one probably the mostrecognised British architect in
history and when you look at hiswork you will understand that
(56:24):
he had actually in some waylooked in depth to the Romans
and the Greeks.
And he actually did depth tothe Romans and the Greeks and he
actually did.
He went on tour around Europe,continental Europe, observing
how these people before him havebuilt buildings.
You know and learned from them,and he uses all of those
(56:50):
techniques in his design.
Now, when you look at somethinglike the Sheldonian Theatre and
the ventilation in theSheldonian Theatre not even the
daylight, which is absolutelyglorious in a space in a country
that is Today- it's sunny, butit's usually overcast, it's
incredible.
And he used all of thetechniques that the Romans and
the Greeks have used before himand made them better.
So when the conversation aboutcopying happens and I do
(57:16):
copyright it's a differentconversation.
But learning from others and,again, bridging that gap between
cultures and bringing themtogether and sharing their
knowledge is absolutelyfundamental.
We grow together.
You know, we don't grow.
Human beings cannot grow ontheir own.
They never could, they neverwill.
(57:38):
Even when the most brutalmiddle of our history, when
there were invasions and I don'tknow when the Romans invaded,
they learned from the othercultures and they actually share
their knowledge with them in avery violent way, but they
actually did it, you, and it'sfascinating, it's, I think.
I think that it's celebratecelebrating this sort of events.
(58:00):
It's important because it's notjust a place where people show
their, their innovations and tryto get a prize or try is.
They're sharing knowledge.
We're coming together.
We're sharing knowledge thateverybody is a winner.
Speaker 2 (58:14):
Yeah, yeah, absolutely.
And you know, you said veryrightly, you have to learn, and
we say also often you have tolearn from your mistakes, but
you can't make all the mistakesyourself in your lifetime, so
you have to learn from otherpeople's mistakes as well,
absolutely.
But at the same time, you alsolearn from the good things that
people do.
So by looking around, youabsorb that and there's no doubt
(58:37):
that you see things and thensomehow, subconsciously, you may
use it in your own designs.
It's logical.
Listen.
I would like to wrap up withasking you is there anything
that you would like or wish thatwe can do in the near future
with AI?
You know, looking ahead to thefuture of our profession, I mean
(59:01):
, you and I are seasonedlighting designers and this
whole AI thing is going todramatically impact on what
we're doing.
You have gone through that insome of your expressions just
now.
Is there something still onyour mind that you think, oh no,
this is really something that Iwould love that we can do soon,
(59:23):
or is maybe something you canalready do, some sort of parting
message that you would have forthe new generation of uh live,
designers of the future?
Speaker 1 (59:35):
I, I'm, I wish, um, generally and this is this is
something that I speak with mydesigners quite a bit which is,
I wish generally that they allowthemselves to play with light
more often than not.
We are lighting designers.
We spend quite a bit of timebehind a computer doing things
(59:56):
for them to enable the play, andthe play seems to be a really,
really small portion of our job.
I enjoyed and I enjoy the mostwhen I'm actually seeing lights
play around me and I'm playingwith them, you know, even
touching them or looking at themor experience them or setting
levels and I wish that AIenables us to do much more of
(01:00:20):
that by doing all of the mundanetasks for us.
You know, just streamline allof the time that we spend in
front of a computer trying toconvey a message that we enable
us to play, to be able to allowus to play with light.
So, if you're a lightingdesigner, you like playing with
(01:00:41):
light and therefore you need tobe playing with light more often
than not, and AI shouldactually simplify that.
Yeah, yeah, yeah, I think thatthat's what I wish for and I
think that, across anyspecialism, I think that if AI
can actually bring that level ofjoy of enabling people to do
(01:01:03):
what they're best at doing,which is, in our case, lighting
design.
That would be brilliant.
I don't want AI to do lightingdesign for me.
It's the opposite.
I want AI to actually make mylighting design better.
You know AI is not going toreplace lighting designers.
They're not.
Nobody will be able to bereplaced when they have that
(01:01:26):
wealth of knowledge in aparticular subject.
So we have developed apps inthe past, in in early 2000s,
which we actually created apicture and people can actually
plot lights onto a building andilluminate.
They're not lighting designers.
They don't know they.
They have an inclination to dothat.
They don't have an inclinationto play with it.
So it's not that an ai willresolve the lighting design.
(01:01:49):
Uh, for them.
You know it's, it's, it's notgoing to happen.
They will always be.
Speaker 2 (01:01:54):
The job of lighting design will exist uh, you, you
brought an interesting topicthere because we didn't discuss
that but the fact that you needhands-on experience to see what
light does um, I think we wealways said in our office don't
specify a light that you youhave never touched or played
with, because you wouldn't know.
(01:02:15):
It's important to really havethat touch and feel and
understanding visually and touchand feel type of thing to
understand that.
So that's something whichbecause certainly, if you talk
about the experience of lighting, you can go into a VR or AR
side of the environment and dosome virtual reality with
(01:02:36):
playing around, but that's notthe same and that's a bit.
The risk that lies ahead isthat people are going to
experiment with lighting in avirtual space, right, rather
than actually feeling it as ahuman being, and I think that
point that you bring up is areally, uh, powerful message I
(01:02:56):
think we need to make sure, as a, as a very small image, yeah,
as a takeaway.
Speaker 1 (01:03:02):
If ai can actually get us out of sitting in front
of a computer to be able to playwith whatever we like to play
in my case, light that's a win,isn't it?
and I think train your eye modelto tell you to get out and play
with light, exactly yeah, justsay, don't worry, uh, I'll deal
(01:03:23):
with the nonsensical bits ofyour work and you go and play
with light and be a betterlighting designer, I think I
think I think that's what I wishfor.
On a last note, I'm exploringthe imagery that AI can generate
, especially.
I have explored in depth thestatic imagery that AI can
(01:03:48):
generate, but the moving imagesis what I'm actually focusing at
the moment.
So, um, that that's, that's a.
That's an advance that I'veseen.
So I've seen ai now actuallyunderstanding image terms, text,
so they can actuallyincorporate text in a way that
they couldn't a few months ago.
The moving, the moving imagesare happening, but they're not
(01:04:11):
there just yet.
So I'm really looking forwardto the next two, three months in
which all of these AIs aregoing to be developed and
probably feed into our business.
Speaker 2 (01:04:20):
Maybe we should have this chat again in a couple of
months and see how things haveevolved.
Speaker 1 (01:04:25):
Yeah, definitely We'll have it in, like Futura.
So basically we'll definitelytalk about these sort of things
with some visual material there.
Speaker 2 (01:04:35):
Well, you know, creative minds like you and me,
we trigger each other every timewe say something and it opens
up another idea or another thingthat we want to talk about.
But I'm going to wrap it up.
Speaker 1 (01:04:46):
Yeah, let's do that.
Speaker 2 (01:04:49):
It's been an absolute pleasure to talk to you, and
thanks for your time and sharingyour insights.
Speaker 1 (01:04:55):
Thank you, Martin.
Thank you for the invitation.
It's a pleasure to me toactually speak with you all the
time.
I don't want AI to do lighting design for me.
It's the opposite.
I want AI to actually make mylighting design better.
You know AI is not going toreplace lighting designers.
They're not.
Nobody will be able to bereplaced when they have that
wealth of knowledge in aparticular subject, and AI is
(00:20):
nothing more than a tool thatlearns.
So if you actually teach themthe wrong things, it will give
you the wrong result.
It's important that we trainour models in the most
appropriate manner and by thepeople that know the subjects.
Any mistakes that you make ontraining your models or your
agents then cannot be erasedfrom their memories.
It feels that you need torestart training something
(00:44):
different in order to get theright result.
All of your work can go quicklyto the bin if you actually make
a mistake in that front.
Speaker 2 (01:02):
Juan Ferrari.
Welcome to Light Talk.
Speaker 1 (01:05):
Thank you.
Thank you, Martin.
It's a pleasure to be here withyou.
Speaker 2 (01:09):
It's a pleasure to actually have conversations with
you in any platform and in thisplatform yeah it has been quite
a while since we last spoke,that's for sure, but I'm having
you here because I want to talkabout AI and the future of
lighting design.
But before we dive into that,give me a little bit of your
background.
A lot of people know you, butthere's also probably some of
(01:30):
our audience that don't know youso good.
If you gave us a bit of yourbackground in terms of where
were you born, what did youstudy, how did you get into the
position you are today?
Speaker 1 (01:42):
It's a very convoluted journey, as most of
the lighting designers in theindustry I I I'm argentinian.
I was born in buenos aires many, many, many years ago I'm not
going to tell you when was that,but many years ago, you need to
trust me there.
Um, I trained as an actor andand then I started directing
(02:02):
theater and then I actuallystarted performing and then I
started getting involved in allof the technical aspects of
theatre, and technical aspectsthat I liked the most was
lighting, and I became kind ofan accomplished theatre lighting
(02:23):
designer To the point that Istarted doing things that were
over and above what I thought Iknew.
So I came here to the UK tostudy theatre lighting design.
I studied at the Royal CentralSchool of Speech and Drama, I
did my course there, I did a BAin theatre lighting and from
(02:44):
there on, while I was atuniversity, I started shifting
my attention to architecturallighting.
I find it fascinating that thenarrative that we had in theatre
and that I have actually livedas an actor and as a director
and as a lighting designer intheatre didn't naturally exist
(03:04):
in architectural lighting.
So I actually started payingattention to that and starting
exploring that, to the pointthat I ended up working in
architectural lighting.
So I got my first job as anarchitectural lighting designer
in a company called Equationthat has now changed hands.
It's the actual equation.
It's the GIA equation, but atthat point, mark Hensman was
(03:28):
actually the director atEquation.
So I started working there.
I met some of my colleagues, mycurrent colleagues.
You had Horley there, inparticular.
Jonathan Rush and John thenmoved to Horley and I moved
shortly after, and since then wehave actually been developing
the lighting team for Hawley,first at the direction of
(03:50):
Dominic Merrick and now underour own direction.
So between me, jonathan andRuth Kelly-Wasket, we are
actually running the lightingdesign department of Hawley,
which is quite a and a very wellrecognized and well established
lighting design department inits own right within a massive,
really big engineering companythat is also part of a bigger,
(04:14):
even bigger multinationalcompany called tetratech.
So that's my role at the moment.
I direct a team of 25 people inlighting design.
They're lovely.
They're really, in my view, themost creative people that I
have actually met, with anenormous understanding of the
technical knowledge that youneed to have in order to deliver
(04:37):
the best quality lightingdesign possible.
So, yeah, that's my rolecurrently and I love talking, so
I do a lot of talks.
No, no no, you're here to talk.
I'm quite curious also, martin.
So that's why I ended upworking and exploring AI, which
(04:58):
is another big story, anotherlong story in its own right.
We'll get to that, we'll get tothat.
Speaker 2 (05:06):
I've got a cheeky question.
That's the fact that you havebeen an actor being of help in
your, in your profession, whenmeeting clients or uh, I think
that I, actually, I, I thinkthat I it helps.
Speaker 1 (05:21):
Communicating anything that you do to
understand your body, tounderstand the way that you
express yourself, to understandthe emotions that you feel while
expressing, and any sort ofregister that you get of your
own tools, of your own physical,emotional tools, is quite
helpful in any environment.
So, yes, it helped me it.
(05:43):
It, to be honest, um, one ofthe most important things for me
is to to be able to communicateproperly, and lighting is a
very difficult.
It looks very easy, but it'svery difficult to communicate in
words, and and I think that my,my theater background have
helped me enormously, and thenthe notion that lighting is a
(06:07):
tool that actually tells storiesand that has a narrative.
All of that piece of work ofmine is based on my experience
in theatre really, so yeah ithas helped me a lot, yes.
Speaker 2 (06:20):
Yeah, I think it's a great combination because in
theatre lighting your focus istechnically on the actors when
you light, but in architecturallighting it's more about
lighting the architecture.
So I think that combinedknowledge I think is great.
Speaker 1 (06:37):
Also in architecture, it's not only lighting the
architecture, it's lighting forpeople using it.
So, although we don'tnecessarily experience our lives
as a play or as a narrative, wedon't feel that our lives are a
story.
We are actually living throughemotions, in the same way that
(06:59):
we are living through emotionsat play.
So it's quite important thatlighting takes into account the
feelings and the journeys thatpeople have within the
architectural spaces, and that'sone of the most important
things for me.
Speaker 2 (07:12):
So let's jump into AI straight away.
At what point of time did youget caught by the AI bug?
Speaker 1 (07:21):
I actually it's quite funny I do a talk about AI and
I actually expose myself quitein a big manner in that
particular talk because I'm auser of AI and user and abuser
of AI.
So I started playing with AI ona journey to Birmingham
actually were we were in the carand and particularly with chat
(07:45):
gpt um.
We were in a car a few yearsago and one of my colleagues
that was traveling with meshared with me this new thing
that have come out called chat,gpt and um, and I asked chad gpt
a particular question about uhlighting and he'd answered it in
a way that I couldn't answer it, even if I was given an hour or
(08:10):
two to respond to thatparticular question.
I went like, wow, this is quiteimpressive because in a
microsecond, all of a sudden Iget an answer that was quite
solid and it was very muchtechnically um and I thought
this is quite interesting.
And so the iald was doing theirenlightening conference and
(08:32):
calling for papers and Iactually composed a paper
completely with ai about ai andlighting, but that without me
knowing a lot about ai, and thepaper got selected.
So all of a sudden I had tolearn about AI in order to
present about AI, which is awhich is a an unfortunate,
fortunate situation.
(08:53):
You know, it was a little bitcheeky from me and that was
quite successful talk because,as I say, I exposed myself into
a point in which I said, look, Idon't know anything about AI, I
just put this into AI and Iprepared a presentation all
through AI.
And then this is my story Ifyou fast forward, this was three
(09:13):
years ago now, almost two and ahalf years ago, and if you fast
forward to today, I am actuallyconstantly training myself on
these new tools and in in aithat are available to us.
I'm playing with all of them, Iuse them on a regular basis.
I I encourage people.
(09:36):
I think that my mission on this, on this particular front, is
to encourage people to use it.
Some people are quite scared.
They have ethical concerns,they have moral concerns, they
have all sorts of concerns, andthe only way of actually making
this tool a less uh, a lesserproblem is using it, you know.
Speaker 2 (09:55):
So I'm encouraging people to use it I'm a bit like
you because I mean, I just gotlate last year something I I got
in touch with it and then, um,I I know if you have seen, but
I'm I'm promoting an ai coursefor for lighting designers,
which I'm doing with an aispecialist which, like you, he's
(10:15):
an architect, he's been in thisfor the last couple of years,
he knows everything of ai, hedreams and and lives ai.
But I I feel also a bit bitreluctant and a bit overwhelmed
by all the possibilities.
And you talked about the toolstwo years ago.
Well, you see what speed theyevolve and develop, so what's
(10:35):
possible today?
But for me, the reason to jumpin is to be the reason to be.
You see, I've got 45 years ofexperience in lighting design,
so I can actually sort of beingthe challenger and the reasoning
behind what AI throws at us andlook at it through the lighting
design expertise that I haveand say, well, is this correct?
(10:57):
Yes, and, like you say, whatcomes back is sometimes amazing,
like really the knowledge.
So, yeah, I think it'simportant to have that balance
and you can't better learn, Ithink, than really diving into
it and embracing it, becausethere's no way we can go around
it.
It's there.
(11:17):
It's there for us too.
It will be like the mobilephone.
It's something that's going tobe there as a new revolution,
and if you're not in it, you'renot going to win it, that's for
sure.
Speaker 1 (11:28):
So talk about tools.
Yeah go ahead.
Yeah, I think that is portant.
I think that you need to beable to use it and to be in it.
There is somethinggenerationally.
It's a really interestingconversation, the generation
conversation.
There is somethinggenerationally.
It's a really interestingconversation, the generation
conversation.
It's like we did within ourlifetimes.
We went from a pencil and apiece of paper to a typewriter I
(11:56):
would put myself in thatcategory To a computer, to the
internet and to AI, and we didthat.
I'm going to disclose my agenow I'm 50.
So within 50 years, I've donethat.
I'm going to disclose my agenow I'm 50.
Yeah, so within 50 years, I'vedone that.
My journey took 50 years.
If you think about it.
Our kids nowadays learned allof that in a period of 10 years.
(12:19):
So imagine how much all of thistool-related relationship that
we have with our profession anyof our professions will evolve
in the next 40 years for them.
What will come, we don't know,but it's going to be definitely
at the pace that we arecurrently exploring it.
So it's going to be magical,you know.
So it's incredible the amountof change that is ahead of us.
(12:45):
So what we need to build inourselves is the possibility of
being flexible and grab thesetools that are given to us.
And today today, yeah, we're in2025.
The tool that we are using, or alot of people are using, is
Chak, gpt or DeepSeq, buttomorrow it will be another one,
and but tomorrow it will beanother one, and the day after
(13:07):
it will be another one.
So what you need to be able isto become permeable to all of
these tools that are coming toyou, to be able to use it, and
the way of really using them ishaving a reason to use them.
So I think that the second bitof an NEI conversation is what
do we want to do with them?
(13:28):
What do we want to do with AI?
Because if you don't have ananswer of what do we want to do
with it, then we cannot use thattool.
Speaker 2 (13:36):
Okay, so tell me, what are we going to do with AI?
Speaker 1 (13:40):
What are we going to do with AI?
What are you doing?
Maybe?
Speaker 2 (13:43):
better.
What are you doing at themoment with AI and what do you
think we should probably bedoing?
Speaker 1 (13:50):
So let's start with a very basic point of what do we
do with AI?
The first thing that we do withAI is try to streamline our
mundane tasks.
You know, those things that wedon't like doing, like writing
emails.
What do I use AI for?
To write my emails.
I turn not to write any moreemails.
I actually tend most of thetime to dictate an email to an
(14:13):
AI.
The AI composes.
I review that email send.
Speaker 2 (14:19):
I still need to correct, you still have to
prompt, so that means you stillneed to write something.
Speaker 1 (14:23):
No, I do that verbally.
It flies into the prompting.
So I do that verbally, I do iton the microphone, so I actually
dictate it to okay all right,okay, okay, yeah, I don't, I
don't, I don't write it.
I actually go um, write an emailto martin class and I want to
actually discuss this nextinterview that we're doing
together.
I want to know the subject thatwe're going to discuss.
(14:44):
Send and the AI does it for me,formats it for me.
When you're working on yoursecond language and English is
my second language.
This is a really, reallytime-saving tool.
It does save me loads of timeand I absolutely adore the fact
(15:04):
of doing that mundane task forme.
I use it quite a bit forresearch, so I build up project
folders within within the AI, myAI of choice, and actually then
say well, look this, thisparticular bot is going to act
as a lighting agent or alighting person, and and then it
has this amount of experienceand actually based all of his
(15:26):
knowledge on this, this, this,this piece of information.
And then I ask say that I wantto research I don't know um the
the of anything, what, what didthe dinosaurs do before they die
?
Basically, I asked the AI totrain as a dinosaur expert.
(15:50):
Then it would tell me back, itwould actually start giving me
facts of what were the lastthings, what was the last supper
from the dinosaurs, and all ofthat, doing research, doing you
do that with.
Speaker 2 (16:04):
NGPT research, or are you using other tools?
Speaker 1 (16:07):
I I use I currently use because of my, of my company
policies.
We are quite restricted of whatwe can actually use or not use
for work.
So I obviously experiment on apersonal computer, but I also,
when I'm working, I have a chatgpt enterprise which is probably
the latest version of chat gptwith a full package around it.
(16:31):
So it has dali, it has saw, ithas other other cameras and
other and other um and otherservices that chat gpt brings.
But it's actually ring fencedso the information is actually
kept confidential.
So I can actually handleconfidential information within
ChatGPT Enterprise withouthaving to share it with the
(16:53):
world, because obviously, as youwould understand, we do things
that are confidential in ourjobs and our clients would be
happy if we disclosed thatinformation into different AI
services.
So I use that quite a lot.
I experiment a lot with thevisual side of AI.
So basically, I use Midjourney,I use Runway, I try to explore,
(17:18):
I'm trying to get to a point inwhich our lives become our
presentation life.
You know that we we have donemany presentations and we we
actually love the movement inthe presentation and we love the
visual engagement in thepresentation, and we live in a
world of instagram now.
So we live in a world thatimages move and flick and move
(17:41):
quite quickly and and I look atour lighting presentations
generally and they're quitestatic.
So I'm trying to get a littlebit more of a different way of
presenting through AI at themoment, which I'm exploring and
I have to admit, martin, I havenot refined this, but I have
actually got to the point that Iactually don't present static
(18:02):
images anymore.
I kind of animate them throughrunway in my presentation.
So I use that quite a bit andit's quite interesting.
The problem with images andmoving images and generating
moving images in AI is that itstarts deforming objects and I
know there are some certainpeople that do it in a marvelous
(18:23):
way.
I'm not one of them, but I'mgetting there.
I'm not a.
I know there are some certainpeople that do it in a marvelous
way.
I'm not one of them, but I'mgetting there.
I'm actually playing around andtrying to get to that sort of
image movement within the AIworld that I'm exploring.
And then there is the use for AI, of which would be probably the
(18:44):
iterative design conversation,when you have to bring a lot of
options, so giving the machinean option and say, look, give me
the lighting design for thisroom.
Or I assume that I want alighting design for this room
with downlights and offer detail, and I want some up lighters
here and there.
Just give me some other optionsand it would quickly give me
(19:06):
options on the way that Iactually could approach the
lighting to that space.
Now I could think about thoseoptions on my own.
Obviously, I've been doinglighting design for so many
years, but sometimes the thequirky nature of ai, the
hallucinations of ai, actuallyopen another dimension, you know
, and give you yes, give you anoption that you have not thought
(19:26):
to and it's kind of ridiculous,but it's lovely at the same
time, you know.
So it's.
It's that, it's another thingthat I use?
Speaker 2 (19:32):
do you challenge your designs as well using ai?
Because I think that's anotheropportunity that we now have is
just feed the design into the aitool and ask it to challenge.
We are currently doing researchand I'm talking about the AI
course that I'm developing atthe moment.
One of the things that we aretrying to figure out is whether
we can input, like a dialectscalculation for argument's sake
(19:59):
and then see whether AI can saycan we improve this, how can we
make this better.
See whether AI can say can weimprove this, how can we make
this better.
But I think it would hold forany sort of design proposal that
you make that you can actuallychallenge your own design
creation.
It's like how could we improvethat or are there any weaknesses
in this?
What works or doesn't work?
(20:20):
Are you doing that as well?
Speaker 1 (20:22):
Yeah, we do that.
So we do that in artificiallighting and in daylighting too.
So we use the basics of what wehave been doing pre-AI in
parametric design with Rhino andGrasshopper in order to, for
example, just actually build upa room and say what's the most
efficient opening that I canhave as a window in terms of, I
(20:43):
don't know, thermal comfort andand um, sorry, um heat gain and
and um and uh and uh, daylight,um, daylight, um, penetration
and and basically we have used,I don't know, grasshopper and
and rhino for many years in ingiving us those numbers, you
(21:04):
have to think that we are anengineering firm, so our numbers
are quite important to us inmany ways and so we have been
playing with them for many years.
So AI actually brings anotherlevel of joy to that and another
level of speed.
And, yes, definitely on theartificial lighting side, if you
don't remember wrong, reluxused to have this tool that you
(21:26):
can give them in a, in arectangular box, not the most
complex of geometries, but in arectangular box.
It would give you the the mostefficient way of actually
placing your luminaires within,within, within your plan, and
and I think that that was pre-ai, you know.
So, yes, we, we have beenplaying around with that.
A challenging design through aiis quite important, in the same
(21:47):
way that doctors challengetheir findings with AI.
You know, you have heard thatdoctors, for example, they have
a scan and they're actuallysearching for I don't know
tumors or things like that, andthey go like, well, have I
actually looked at this x-rayproperly?
And the AI will actually say,well, you haven't looked at this
bit here.
Actually you might want to lookat that, because that's kind of
(22:10):
not very common.
But in order to actually get tothat point, martin, as you would
know, you have to train yourAIs, and that's a critical thing
.
An AI is nothing more than atool that learns, so if you
actually teach them the wrongthings, it will give you the
wrong results, in the same waythat if you input the wrong
(22:30):
parameters into a lightingcalculation, you will get the
wrong results and it will bemeaningless.
So it's important that we trainour models in the most
appropriate manner and by thepeople that know the subject.
So any mistakes?
What I do find out is that anymistakes that you make on
training your models yeah, oryour agents then cannot be
(22:54):
erased from their memories andit keeps on repeating and it
feels that you need to restarttraining something different in
order to get the right results,so all of your work can go
quickly to the bin if youactually make a mistake.
On that front, I think thatdata sets are quite critical, so
(23:17):
basically, having the rightdata to inform these AI agents
is absolutely a must.
Speaker 2 (23:25):
You mentioned training, so are you the only
one in the company doing that?
I would assume that you're notthe only one and that you have
your own team on this.
I can also imagine that you Idon't know whether you have like
an office model and a personalmodel, because you can train, I,
on, on, on on your personal uhdata that you want, uh, whatever
(23:49):
I I tool to follow.
But if you use an enterpriseversion, I don't have an
enterprise version, I have justthe pro version.
But yeah, um, you, you train, I, I'm trying, I'm just, I'm just
a starter, so I'm I'm trainingit all myself.
I can imagine that if you havean I don't know whether the
enterprise works as a collectiveso that multiple people can use
(24:13):
the same tool, so the trainingis more.
Speaker 1 (24:17):
You can create a bot and then share it with other
people.
So, yeah, you can actuallytrain an agent or a bot and then
that share it with other peoplewithin enterprise.
You can do that.
You can do that in pro.
I can actually shareinformation with you that I want
to share with you, or a botthat I want to share with you.
That that shouldn't be aproblem.
Yeah, they are compatible so inthat respect.
Speaker 2 (24:40):
So I can actually do things from pro to pro, can you
maybe explain a few of thetraining things that you're
doing?
How do you train that to doexactly what you want it to do,
or to make it machine learn whatyou want the AI tool to become?
You talk about agent.
(25:01):
Are you really already creatingan agent or is it just a way?
I mean, people talk about agentas an autonomous tool, but I
would imagine in this case,you're talking about an agent as
something that you have trained.
Speaker 1 (25:17):
Yeah, more starting with a bot concept and then
trying to get it to go to anagent.
But the bot concept isbasically you have your GPTs.
Yeah, just create your own GPT.
It's exactly that.
That's what I'm trying to say.
What we do is we create our ownGPT.
A GPT, for example, can be, saythat you wanted to create room
(25:39):
data sheets, sheets.
So what you need to embed intothe bot, into the GPT, is the
knowledge of where they need toactually grab the parameters in
order to create a room datasheet.
So you will actually connectthe knowledge of the AI to the
(26:03):
standards that you need toactually comply with.
And you will feed all of thatand you'll say, look, on the
room data sheet level, I need toget this parameter.
So, basically, I don't knowilluminance levels, glare, yeah,
all of the parameters that weuse in lighting.
And from there on, the machinewill actually start
(26:27):
understanding how to get you tothat point.
So, basically, then you willupload a drawing.
It will say you have I don'tknow you have.
Say that you're in a hospital,you have a clean room, you have
a treatment room, you have a Idon't know an operation theatre.
Yeah, then it would justtabulate in a table very quickly
(26:49):
all of the levels and all ofthe parameters that you need to
hit within each one of thoserooms.
That's an enormous time-savingbot, for example, for us.
So basically we use that on aregular basis.
Speaker 2 (27:04):
But I can imagine you can also create your spec
sheets and other documentationas templates that you can just
call back if you want to.
Speaker 1 (27:19):
We are quite so.
We have actually done aswimline diagram of all of our
processes.
So this is outside of AI.
So we've done lighting design.
What do we do From the momentsomebody tells us we have a job
for you?
Speaker 2 (27:39):
We would like to discuss this.
Speaker 1 (27:41):
What do we do?
What sort of information?
What sort of information goesout in what sort of program?
Who gets involved in it?
So we have actually done aswimline diagram of all of our
processes and we are trying,against each one of those
processes, different AI tools.
Now we haven't yet developed atool, an AI tool that assists us
(28:04):
with each one of thoseprocesses.
We've been miles away from someof them, which are the ones
that we would really like tohave.
So, for example, I would verymuch like AI to assist us with
the Revit integration of ourdesign.
So if we had a model, if,ideally, we have a model from
day one how AI can actuallystart populating things in a
(28:29):
logical manner for us, becausethat's a really time consuming
exercise for anybody at themoment and we're trying that.
That relationship between thecalculation program and Revit
and how they link together.
We seem to be modeling andmodeling, and modeling and
modeling, and AI doesn't seem tobe able to help us.
So we do a model for acalculation and we populate a
(28:51):
model like if it was a drawing,which is really uncomfortable.
And why are we not using thesame model?
So that streamlining all ofthose processes and getting AI
to help us with all of thoseprocesses would be the future.
That's what we're exploring.
But in essence, on thatstreamline diagram that I was
talking to you about, with allof our processes, what we are
(29:12):
testing is against each one ofthose processes and what can we
use and what technology can weuse and what tools can we use in
order to actually make themmore efficient and better?
Speaker 2 (29:23):
And I think the thing is that today it might not yet
be possible, but tomorrow itwill be.
I mean, it's developing andevolving so fast that, um, it's
whatever we have in mind that wethink will be able uh, in in in
the near future will happen.
I'm pretty sure of that.
Um, it's just a matter, likeyou say, you need, you need to
(29:43):
get your hands dirty and andplay with it and explore, but
then you have also, I mean,exploring things within your
enterprise platform is one thing, but then exporting it to a
client is another thing.
So I would imagine that youhave some checks and balances
(30:03):
there in how, what is it thatyou can use towards your client?
In what is it that you can usetowards your client?
You probably need to excerptthings and maybe some you don't
want the client to have and somethings you might think are
private to you.
Speaker 1 (30:20):
So how is that, martin?
In that respect, we're quitetransparent with our clients.
We have systems in place tomake sure that all of our
information, in the same waythat you have IT systems, make
sure that all of our informationdoesn't get leaked and doesn't
get compromised, and thereforethat's why we work with
(30:43):
enterprise, which maybe is alesser it's not the pro version,
it's probably a step behind butwhat gives us is a ring fence
of the information and makingsure that everything is secured
within our own service.
So that is quite critical forus.
Speaker 2 (30:58):
So and then you know what I meant I meant to upon is
that you, if you do it on anindividual basis, you control
what you put out there, but ifyou have a team, you still need
to have a unified company uhrestricted or company managed uh
output and and the bigger yourteam, the more complex that
(31:21):
might be, because people may goin different directions.
So I mean standardizing the wayyou present to your client.
That was more what I meant.
Speaker 1 (31:30):
Yeah, well, it's part of the quality management that
we actually do.
So any company, any largecompany, you would know.
We have quality managementrules and basic quality
management procedures that weneed to follow, and we actually
do the same with AI.
So AI is no different to anyother things that we do.
We had to.
(31:53):
There is something it's a bitmore complicated to control In a
world that is never evolvingand you want people to explore
tools.
It's very difficult to do itwithin a multinational company.
So that's where you actuallyhave dedicated exploration and
research machines and then youhave your day-to-day machine in
(32:16):
which you can do your job in asecure manner.
But that comes onto theprofessional.
The way that we handleinformation needs to be very,
very much controlled, because ifnot, we would be making things
that we don't want to leak.
So, uh, but, but.
But at the same time, you wantto encourage your design team
and your people to explorethings.
What?
(32:36):
What actually surprised me andI'm going to go back quickly to
one of your questions I don'tthink I answered which is you
said to me did you, does?
Your whole team works with aiand we work at very different
levels, you know, and in verydifferent ways and I encourage
them to work with AI.
I actually go like please usethem.
We have loads of options withinthe suites that we have within
(32:59):
Holi for them to actually playwith, but sometimes they don't
find it's like with everything.
Sometimes a tool is not usefulto them and they actually might
choose to do it in a differentway and without the assistance
of an ai.
So that's, that's a choice, youknow, and and um, I don't think
you can push people into ai.
I think it's a pity if people donot go into the eye because,
(33:22):
exactly, martin, it's the samething that happened with I don't
know autocad.
You would actually said topeople at some point there were
people drafting drawings anddoing lines and working on it,
and, and, and those peoplethought that they could keep on
drawing by hand and some of themwere reluctant to actually
(33:44):
engage with AutoCAD and at somepoint nobody was drawing anymore
and everybody was drawing inCAD.
So it happens with every singletool that comes into the market
.
People would have said the samething from the internet, from
Photoshop, from everything, thecalculation programs.
Speaker 2 (34:00):
People used to do lighting calculations by hand.
Yeah, some people are morecomfortable than others.
I mean some are sort ofnaturally flowing into it.
Others I mean I'm literally oldgeneration, I started when we
didn't have are more comfortablethan others.
I mean some are sort ofnaturally flowing into it.
Others, I mean I'm literallyold generation, I started when
we even didn't have computers.
So you know this my first, myfirst encounter with the
computer was a huge mainframethat was sitting in a big room
(34:20):
where you had to do punch cardsand things like that for
lighting calculations.
But that's well, okay, we talkI'm.
I'm slightly older than you.
I won't disclose my age, but 51.
So how many people would yousay are working on AI now,
percentage-wise?
Because of course I imagine youstill have a bit of a dedicated
(34:44):
team.
Yes, you can encourageeverybody, maybe even on email
level, but even actual designproduction, that might be a
different sort of approach.
Speaker 1 (34:57):
How many people from my team, or how many people
within Holi, or how many peoplein the world?
Speaker 2 (35:03):
Okay, well, what is within your?
That's a really interestingquestion.
I actually Just expand.
Well, what is within your Of myteam?
That's a really interestingquestion.
Speaker 1 (35:11):
I actually I don't have a.
I have to guess the answer ofall of the three questions.
Obviously, the world one is Idon't know it would be of the 7
billion people that we are.
I don't know how many peopleyou say I guess that a lot, but
on my team.
So I have 25 strong in our teamand I think on a daily basis, I
(35:36):
would say that probably 70% ofour team engages with AI.
Pretty sure that everybody hasengaged with AI in one way or
another within our team and somepeople do more regularly than
others.
As I say, some of us use it allthe time.
Speaker 2 (35:56):
Yeah, I mean it's a bit addictive, let's face it.
Once you get onto it, it's likeit's hard not to use it.
Speaker 1 (36:05):
Absolutely, absolutely.
I use it as a search engine.
I don't search anything else inGoogle anymore, I just search
in ChatGPT.
I find it easier.
I find that the questionseasier, the answers easier.
Speaker 2 (36:20):
You know, even if you use Google search, you get into
Gemini, it comes up with.
The first thing that comes upis a Gemini answer.
Right yeah?
Speaker 1 (36:28):
I don't even like the expression of Google anymore.
Basically, google gives me Idon't know 100 results with
three or four ads in the middle,and I've got to go.
I need a phone number and I goto Chachipiti.
I dictate to Chachipiti I needMartin Klaassen's phone number
and it will just give me yourphone number.
Not a lot of ads and a lot ofyou know, soon you can even use
(36:50):
my voice.
Speaker 2 (36:53):
I'm currently cloning my voice.
I have the first clone and Iwant to use it to do an e-book,
so that you know my books.
I can read them, but I don'twant to read them.
So I'll get my cloned voice toread my books, that's very good.
Speaker 1 (37:08):
Writing a book is one thing, you know.
I find that fascinating.
What you're telling me is soenriching, Martin, because it's
endless the amount ofopportunity.
I've never thought about it.
Actually, I particularly have aproblem with reading, because I
read very badly.
Imagine when you're reading ona second language.
My eyes go completely crossed.
(37:30):
But I adore the narrative of abook, so I adore the stories
behind the book.
Now, what you just told meopened my eyes, like that, you
know, because it's great and Inever thought about oh, why
don't I actually train my AI tohave my voice and then give them
the books and read them back tome?
That's perfect.
It's an audio book with my ownvoice, which is great.
(37:51):
Yeah, yeah, yeah.
Speaker 2 (37:53):
Well, soon you can read my book or listen to my
book with my voice, exactly,exactly, remind me, I'm like you
.
I followed an AI course earlierthis year that really opened my
eyes.
It was supposed to be athree-hour course.
It ran out to nearly four hoursand the opportunities and the
(38:18):
things that were possible, thatblew my mind away.
And that's when I dived head into develop this AI course,
because I think people need tounderstand what's possible and
get over the threshold of angstand the threshold of fear of
what it all is and just see itfor what it is.
And then, of course, you needto make up your mind of what's
(38:41):
good and bad about it, and notforgetting that it is an
assistant and you really need tomake sure that you keep the
creative control and thecreative mastership about what
it all creates Right.
But yeah, I think theopportunities are endless.
Speaker 1 (39:01):
So I am doing a course called AI for Business
Value and it's actually a mixbetween AI and a business course
, which is quite interesting.
It's on the right, it's anapprenticeship by Multiverse,
which is a very basic course onboth business and AI.
But one of the things that isquite good about that course is,
(39:24):
first, that everything in thecourse, within reasons and
within the right parameters andthe right way of doing it,
should be and could be run by ai.
You know so.
There is no.
There is no.
You need to read this book.
It's you.
Do whatever you want.
You.
Give me the solution howeveryou want.
You, use the hour however youwant you, even if you want to
play back everything that aisays to you.
(39:45):
Obviously, there is a risk indoing that and we all know it.
That's completely acceptable.
So there is no.
This is a task that you shouldbe doing on your own without AI.
Ai is there.
It's like if nobody's notallowing you to use a pen, you
know, here's a pen or a pencil.
Ai is a pencil.
Use it.
And one other thing that welearned is that notion of making
(40:10):
sure that the data sets arecorrect and that you're the
reasons why you're using AI,that you have a reason to use it
.
What do you want to do?
What bit of the lighting designprocess do you want to
streamline and then find theright tools to do that.
Speaker 2 (40:29):
It's quite an interesting course well, I I
started, I'm still lovesketching and I, I I'm on the I
don't know if you have the samething, the remarkable tablet
where I can still draw, but it's, it's now digital, right.
Yeah, you know the mark.
So, and now I've started to doa sketch and then input it into
(40:50):
an AI tool and ask it to developit further, just from my sketch
.
Yeah, so there's also endlessopportunities.
Speaker 1 (40:58):
It's great, it's absolutely great.
So, for example, one of thethings that you know when you go
into a meeting, you take notes.
Yeah, yeah, and sometimes youdo take notes by hand and
sometimes you type them in.
I tend to take notes by handwhen I'm in a meeting, but
(41:18):
before I actually used to lookat my notes and type them back
and play them back to I don'tknow a client or my colleague or
whatever.
Yeah, I now take a picture ofmy notes, upload it into chat
gpt in this case and I extractthe text and I have them all
typed in.
Speaker 2 (41:33):
Yeah, Well, you can also ask to make an audio
version of it so that people canlisten to it.
They don't even have to.
Yeah.
Speaker 1 (41:41):
I haven't done that but, yes, we can yeah.
Speaker 2 (41:43):
Yeah, yeah, yeah, those are all of them.
I interviewed someone recentlyand it was also about an hour
interview, but I asked NotebookLM to do a short version of it,
like less than 50 minutessummary, and then you get two
people talking to each other and, in a very, very concise and
(42:05):
great way, you can just listento the summary.
In that way, you don't evenhave to look back.
I wanted to ask you somethingabout copyright and intellectual
property, because obviously, asdesigners and creators, we
always look at that element.
How do you see that within theworld of AI?
Speaker 1 (42:27):
It's quite a complex subject, isn't it?
Because AI in Europe, there'sthe EU directives on AI at the
moment, so there is, there ismeaningful legislation, but
there isn't, there isn't in theUK at the moment.
So it that it's, it's.
We're evolving on all of thatsubject, you know, because I
(42:49):
think it's quite critical thatthere is a degree of ownership,
especially when it comes tocreative material.
I think there is something inthere that needs to be
legislated and I think it willbe legislated properly.
If you ask me, I'm quite carefulwith that, because when you're
(43:11):
extracting and you're generating, I try to generate my own work.
You know, I've actually tried,but I'm sure that the model have
learned from other people and II don't.
I wouldn't be able to tell ifI'm actually infringing a
copyright or not on a particulartext, because I wouldn't know
what the combination of wordsthat other people have used to
(43:31):
actually talk about a particularsubject.
So it very much everything thatgoes out and I stress this out
everything that goes out thatwas created by AI needs to be
stated, as with AI assistance.
You know you need to.
Actually we need to write intoour, into a feedback.
(43:51):
Yeah, you need to declare it.
I know you need to.
Actually we need to write intoour, into a fee.
Yeah, you need to declare, youneed to declare, yeah into into
the, into the contracts that wego into, that we are going to
use ai in that work.
Speaker 2 (44:04):
It's absolutely fundamental that that is the
case so your case is actually inyour, in your contract.
You already put in yourcontract.
Speaker 1 (44:12):
We state that we would be or we could potentially
be using.
Yes, we just need to startdoing that because, if not, it
would be not being very clear onthe way that we are working.
No, I think we need to behonest and transparent about
these things.
Speaker 2 (44:29):
Yeah, I agree, and now is the time to start
mentioning that, I think,because most people are starting
to get familiar with it and youknow there's also ways to
potentially protect your work incase of non-payment, like now.
We could potentially I don'tknow whether you, but it is
(44:50):
something that's being exploredby putting all your deliverables
in the cloud with asubscription link and if you
don't pay your subscription, youdon't have access.
You cannot download it, you canonly work in there.
You can see everything, but ifa client, for argument's sake,
would not pay, then you can'taccess the work anymore.
(45:13):
We all experience situationswhere clients are tough
paymasters and sometimes theywant everything for free, or
they want everything and theydon't want to pay, or they are
very difficult in payments.
But potentially, through AI andto this kind of digital and to
this kind of digital cloudsituations, we could better
(45:34):
manage that our whole paymentschedules and all that.
Speaker 1 (45:41):
Yeah, I suppose that's probably something that
other companies would startactually exploring.
I wouldn't be surprised ifcompanies that have to do with
legal, with legal, I don't knowlaw companies, law firms are
(46:03):
actually already looking intohow that works and how they can
bring a service on the back ofthat.
Yeah, yes, in terms ofcopyright, just to wrap up the
idea, it's very simple, Martin.
Everything that I send is myresponsibility, regardless of
how I actually create it or not.
You know, whatever I used tocreate it, if I press send on an
(46:25):
email, nobody's pressing sendfor me on the email.
Yeah, so if I created an emailwith AI that says all sorts of
horrible things.
It's me sending it.
So the one thing that we needto be careful in the same way
that we are careful withanything that we do is that when
we actually press send, that isa piece of information that we
agree with.
(46:45):
There is no point of blaming.
There's enough law casesalready.
Speaker 2 (46:52):
It's part of your already.
It's part of yourprofessionalism, part of your
integrity, you know, andtransparency.
I think in order to do business, you need to be in integrity,
also for the longevity of yourprofession and your
relationships with the client.
I think, yeah, and also it'svery dangerous.
Speaker 1 (47:12):
It has.
There are many law casesalready of people that actually
say no, well, sorry, I wasactually doing this with an ai.
I was well, who cares?
Speaker 2 (47:18):
you know, it's your, it's your responsibility, uh,
yeah, yeah, exactly, it's likesay look, I didn't draft this,
because I was drafting it in inin autocad or in revit, or I was
actually creating this model inrevit and I'm sorry, it's not
it's not real because it's yourresponsibility Exactly, and it
takes years and years to buildup trust and respect, but it can
(47:38):
be destroyed in minutes, so youreally have to be very careful
on how you treat yourrelationships in that respect.
I want to make a littlesidestep to Lux Futurum.
You are a guest speaker on theLux Futurum.
You are a guest speaker on theLux Futurum platform.
You will be doing one of thepromo tours in China, and thank
(48:01):
you for that.
What does I mean?
You know a little bit more aboutLux Futurum now.
It's a platform and arecognition program that aims to
recognize future ideas, futureconcepts, future projects,
products, things that sort ofembody the spirit of the future,
new innovations, and we reallywant to bring that foreground
(48:25):
and give people the ability andopportunity to promote and share
their ideas.
And not only that the winnersthat are being shortlisted are
then invited to the final eventat the end of the year to
actually present, because wefeel that the learning
experience and the knowledgesharing about why something was
(48:48):
a selected winner is asimportant as just selecting it,
because we want to share thatinformation.
What is it that made that sospecial?
Um, I just want to hear a few,a few insights from from your,
from your point of view, uh,what you think about luxury
juror and and why you're so uhnice for us to support it oh,
(49:11):
first I want to actually thankyou for the invitation.
Speaker 1 (49:15):
I think supporting any platforms that are actually
looking at innovation is key toany development of any
profession, not only in lighting.
I also really believe in thecrossbreed of specialisms in
terms of how they integrate, sothis sort of innovation
(49:38):
platforms will allow people tostart thinking a little bit
outside the box, and not only inlighting terms, but in lighting
terms connected with otherelements and how they all, and I
think that that brings anotherlevel of innovation and design
and development.
(49:58):
I've heard some reallyinteresting talks about people
that were doing innovation innot only lighting design terms,
but in artificial intelligenceterms in the 1970s, intelligence
terms in the 1970s, and ifthose people would have had
these platforms at thatparticular point, maybe we would
(50:24):
have actually engaged with AImuch earlier and we wouldn't
have the winters of AI, as theycall them, where people just say
, well, it can't be bothered byAI anymore and they'd stop
researching and they'd stopdeveloping because we lost 20
years in doing that, you know.
And now, 20 years losing 20years or not losing 20 years,
who cares?
But I thought that we wouldprobably have a different way of
(50:46):
working already if AI wouldhave been developed around the
1970s and 80s, instead ofactually being stopped and
frozen on an AI winter.
You know, very likely thoseplatforms, these platforms like
Futura, are the places forpeople to speak, to amplify what
(51:07):
they're doing, to capture moreimaginations, to capture more
funding.
To capture more imaginations,to capture more funding, to
capture more track speed, and Ithink that's why I support them,
you know, and that's why I liketo participate in them.
They enrich me, martin, that'sthe thing, you know.
I listen to people innovatingspeak and they enrich me my
(51:31):
persona.
I love that.
I actually have to say thatit's great because, in the same
way that this conversation hasenriched me with the, you know,
when we're talking about yourbook, that you're reading your
book to yourself.
It's great, you know it'sfantastic.
It is fantastic.
Speaker 2 (51:48):
You'll be cloning your voice very soon.
Speaker 1 (51:51):
Definitely.
Why not?
Why not it's?
used in some medical you knowthere are some really bad health
conditions that actually peoplelose their voices and the
people record their voices andbe able to play them back.
How good is that, you know how?
How really good is that andactually to be using it without
(52:14):
my need.
Is not that because I don'thave a voice?
It's because I really don'tlike the process of reading.
I love reading, but I don't.
I find it painful in my eyesand in my head.
Yeah, yeah, yeah, and it'sgreat that somebody, or my own
voice, reads it back to me, youknow, or my own voice reads it
back to me.
Speaker 2 (52:32):
You know, interestingly, in last year's
submissions for Lux Futurumthere was hardly any AI-related
submission Quite surprising, butI would suspect that this year,
with AI taking such a flight,that we would have more
AI-related entries.
I hope so.
(52:53):
I haven't seen any yet with AI,but I would imagine and this
year we're also specificallyasking all the entrants to list
and declare AI tools that theymay have used in the submission,
because I think, for clarityand also for educational
purposes, it's important tounderstand what is humanly
(53:15):
created and what is created withthe help of AI tools.
Yeah.
Speaker 1 (53:24):
I think it's actually good to actually I think it's
necessary to declare what you'reusing as a tool.
The reality is that I wouldguess that most of the people
have used AI last year.
They just didn't declare it or,in some cases, they even hide
it because they think there is anotion and this is a really
(53:48):
interesting bit of aconversation which is a notion
that everybody that is using AIis cheating.
We need to move away from that.
Everybody that is using AI isusing a tool that is powering
and amplifying what they'redoing.
They're not cheating, you know.
Speaker 2 (54:08):
I think, if you declare that sort of idea of
cheating, because you know I'mhonest, I'm using this.
I'm still the creator-in-chief,but I'm using this tool to help
me.
Another point that I wanted tohighlight is that one of the
goals of Lux Futurum is also tobridge East and West.
You know China is thebirthplace of.
(54:30):
You know where mostmanufacturing of the lighting
technology is being done.
Most of our light is all comingfrom china.
90 is all made there and, ifnot, the full feature, the
components are coming from there.
Um, and there is notion also,like you said, that oh, china,
they are copiers and they're notreally, um, you know, at the
(54:51):
forefront of everything.
But that has changeddramatically.
China is very much aninnovative country in that term.
They, in some aspects, areahead of the Western world in
terms of digital integration,use of AI in many applications,
(55:13):
and we felt that there's thisapprehension about China, for, I
mean, I've been coming to Chinafor many, many years, so I
don't have that.
I know these lovely people andthey have moved things forward
quite dramatically, and I thinkthe ability to connect East with
West through this platform isalso what drove us to develop
(55:36):
this.
I would imagine that that'ssomething that you can see.
Speaker 1 (55:42):
Learning from others, from other cultures, from other
people, is absolutely criticalfor any big shift and big
development.
I always use the same exampleChristopher Wren, which is the
architect of St Paul's Cathedraland the Sheldonian Theatre in
(56:03):
Oxford.
One of them did the master planfor London after the Great Fire
.
He is one probably the mostrecognised British architect in
history and when you look at hiswork you will understand that
(56:24):
he had actually in some waylooked in depth to the Romans
and the Greeks.
And he actually did depth tothe Romans and the Greeks and he
actually did.
He went on tour around Europe,continental Europe, observing
how these people before him havebuilt buildings.
You know and learned from them,and he uses all of those
(56:50):
techniques in his design.
Now, when you look at somethinglike the Sheldonian Theatre and
the ventilation in theSheldonian Theatre not even the
daylight, which is absolutelyglorious in a space in a country
that is Today- it's sunny, butit's usually overcast, it's
incredible.
And he used all of thetechniques that the Romans and
the Greeks have used before himand made them better.
So when the conversation aboutcopying happens and I do
(57:16):
copyright it's a differentconversation.
But learning from others and,again, bridging that gap between
cultures and bringing themtogether and sharing their
knowledge is absolutelyfundamental.
We grow together.
You know, we don't grow.
Human beings cannot grow ontheir own.
They never could, they neverwill.
(57:38):
Even when the most brutalmiddle of our history, when
there were invasions and I don'tknow when the Romans invaded,
they learned from the othercultures and they actually share
their knowledge with them in avery violent way, but they
actually did it, you, and it'sfascinating, it's, I think.
I think that it's celebratecelebrating this sort of events.
(58:00):
It's important because it's notjust a place where people show
their, their innovations and tryto get a prize or try is.
They're sharing knowledge.
We're coming together.
We're sharing knowledge thateverybody is a winner.
Speaker 2 (58:14):
Yeah, yeah, absolutely.
And you know, you said veryrightly, you have to learn, and
we say also often you have tolearn from your mistakes, but
you can't make all the mistakesyourself in your lifetime, so
you have to learn from otherpeople's mistakes as well,
absolutely.
But at the same time, you alsolearn from the good things that
people do.
So by looking around, youabsorb that and there's no doubt
(58:37):
that you see things and thensomehow, subconsciously, you may
use it in your own designs.
It's logical.
Listen.
I would like to wrap up withasking you is there anything
that you would like or wish thatwe can do in the near future
with AI?
You know, looking ahead to thefuture of our profession, I mean
(59:01):
, you and I are seasonedlighting designers and this
whole AI thing is going todramatically impact on what
we're doing.
You have gone through that insome of your expressions just
now.
Is there something still onyour mind that you think, oh no,
this is really something that Iwould love that we can do soon,
(59:23):
or is maybe something you canalready do, some sort of parting
message that you would have forthe new generation of uh live,
designers of the future?
Speaker 1 (59:35):
I, I'm, I wish, um, generally and this is this is
something that I speak with mydesigners quite a bit which is,
I wish generally that they allowthemselves to play with light
more often than not.
We are lighting designers.
We spend quite a bit of timebehind a computer doing things
(59:56):
for them to enable the play, andthe play seems to be a really,
really small portion of our job.
I enjoyed and I enjoy the mostwhen I'm actually seeing lights
play around me and I'm playingwith them, you know, even
touching them or looking at themor experience them or setting
levels and I wish that AIenables us to do much more of
(01:00:20):
that by doing all of the mundanetasks for us.
You know, just streamline allof the time that we spend in
front of a computer trying toconvey a message that we enable
us to play, to be able to allowus to play with light.
So, if you're a lightingdesigner, you like playing with
(01:00:41):
light and therefore you need tobe playing with light more often
than not, and AI shouldactually simplify that.
Yeah, yeah, yeah, I think thatthat's what I wish for and I
think that, across anyspecialism, I think that if AI
can actually bring that level ofjoy of enabling people to do
(01:01:03):
what they're best at doing,which is, in our case, lighting
design.
That would be brilliant.
I don't want AI to do lightingdesign for me.
It's the opposite.
I want AI to actually make mylighting design better.
You know AI is not going toreplace lighting designers.
They're not.
Nobody will be able to bereplaced when they have that
(01:01:26):
wealth of knowledge in aparticular subject.
So we have developed apps inthe past, in in early 2000s,
which we actually created apicture and people can actually
plot lights onto a building andilluminate.
They're not lighting designers.
They don't know they.
They have an inclination to dothat.
They don't have an inclinationto play with it.
So it's not that an ai willresolve the lighting design.
(01:01:49):
Uh, for them.
You know it's, it's, it's notgoing to happen.
They will always be.
Speaker 2 (01:01:54):
The job of lighting design will exist uh, you, you
brought an interesting topicthere because we didn't discuss
that but the fact that you needhands-on experience to see what
light does um, I think we wealways said in our office don't
specify a light that you youhave never touched or played
with, because you wouldn't know.
(01:02:15):
It's important to really havethat touch and feel and
understanding visually and touchand feel type of thing to
understand that.
So that's something whichbecause certainly, if you talk
about the experience of lighting, you can go into a VR or AR
side of the environment and dosome virtual reality with
(01:02:36):
playing around, but that's notthe same and that's a bit.
The risk that lies ahead isthat people are going to
experiment with lighting in avirtual space, right, rather
than actually feeling it as ahuman being, and I think that
point that you bring up is areally, uh, powerful message I
(01:02:56):
think we need to make sure, as a, as a very small image, yeah,
as a takeaway.
Speaker 1 (01:03:02):
If ai can actually get us out of sitting in front
of a computer to be able to playwith whatever we like to play
in my case, light that's a win,isn't it?
and I think train your eye modelto tell you to get out and play
with light, exactly yeah, justsay, don't worry, uh, I'll deal
(01:03:23):
with the nonsensical bits ofyour work and you go and play
with light and be a betterlighting designer, I think I
think I think that's what I wishfor.
On a last note, I'm exploringthe imagery that AI can generate
, especially.
I have explored in depth thestatic imagery that AI can
(01:03:48):
generate, but the moving imagesis what I'm actually focusing at
the moment.
So, um, that that's, that's a.
That's an advance that I'veseen.
So I've seen ai now actuallyunderstanding image terms, text,
so they can actuallyincorporate text in a way that
they couldn't a few months ago.
The moving, the moving imagesare happening, but they're not
(01:04:11):
there just yet.
So I'm really looking forwardto the next two, three months in
which all of these AIs aregoing to be developed and
probably feed into our business.
Speaker 2 (01:04:20):
Maybe we should have this chat again in a couple of
months and see how things haveevolved.
Speaker 1 (01:04:25):
Yeah, definitely We'll have it in, like Futura.
So basically we'll definitelytalk about these sort of things
with some visual material there.
Speaker 2 (01:04:35):
Well, you know, creative minds like you and me,
we trigger each other every timewe say something and it opens
up another idea or another thingthat we want to talk about.
But I'm going to wrap it up.
Speaker 1 (01:04:46):
Yeah, let's do that.
Speaker 2 (01:04:49):
It's been an absolute pleasure to talk to you, and
thanks for your time and sharingyour insights.
Speaker 1 (01:04:55):
Thank you, Martin.
Thank you for the invitation.
It's a pleasure to me toactually speak with you all the
time.
Popular Podcasts
Stuff You Should Know
If you've ever wanted to know about champagne, satanism, the Stonewall Uprising, chaos theory, LSD, El Nino, true crime and Rosa Parks, then look no further. Josh and Chuck have you covered.
Dateline NBC
Current and classic episodes, featuring compelling true-crime mysteries, powerful documentaries and in-depth investigations. Special Summer Offer: Exclusively on Apple Podcasts, try our Dateline Premium subscription completely free for one month! With Dateline Premium, you get every episode ad-free plus exclusive bonus content.
On Purpose with Jay Shetty
I’m Jay Shetty host of On Purpose the worlds #1 Mental Health podcast and I’m so grateful you found us. I started this podcast 5 years ago to invite you into conversations and workshops that are designed to help make you happier, healthier and more healed. I believe that when you (yes you) feel seen, heard and understood you’re able to deal with relationship struggles, work challenges and life’s ups and downs with more ease and grace. I interview experts, celebrities, thought leaders and athletes so that we can grow our mindset, build better habits and uncover a side of them we’ve never seen before. New episodes every Monday and Friday. Your support means the world to me and I don’t take it for granted — click the follow button and leave a review to help us spread the love with On Purpose. I can’t wait for you to listen to your first or 500th episode!