Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:00):
Phil, then welcome to
Light Talk.
Speaker 2 (00:03):
Thank you very much.
Good to be with you, Martin.
Speaker 3 (00:06):
Yes, it's an absolute
pleasure being here today and
it's actually a real honor to behonest with you.
It's great.
Speaker 1 (00:12):
Thank you, thank you.
Well, I would like to startwith a bigger picture, just to
set the scene a little bit tounderstand and give you also the
chance to explain what BHASchool of Lighting is about,
what courses you are teaching,what certificates people can get
, and, assuming you havestudents from all over the world
, maybe you can let the audienceknow a little bit what the BHA
(00:33):
School is all about.
Speaker 2 (00:36):
You know, BHA School
of Lighting.
The company only started, wasregistered in 2017, we had
(01:00):
started previously under ourlighting design company Banner,
back in 2013.
But then it was in its infancy.
It was very different.
It was a correspondence courserather than being an online
course.
So when we discovered theonline software that was
available, that's when weregistered the company In 2017,
(01:29):
we began operations under thebanner of that company and have
done so ever since, and thecourses, of course, have been
vastly changed and enhanced, sothey are up to the moment,
accurate and dealing with allthe latest technologies.
So that's quite excitingbecause it's constantly on the
(01:53):
move, so you're constantlyupdating your material.
Speaker 1 (02:02):
So when you say
software, what is the technology
?
You're using a specifictechnology to teach.
Speaker 2 (02:06):
We are using
technology known as Moodle.
It originated in Australia.
Speaker 1 (02:14):
Wow.
Speaker 2 (02:15):
Yeah, it's now
worldwide and it's fantastic
software.
It's just got so much to offer.
I'm talking in the teachingenvironment, the educational
environment.
Speaker 1 (02:30):
Right.
Speaker 2 (02:32):
So it's really really
good we can embed video content
.
There's just so much we can dowith it.
Speaker 1 (02:41):
What courses do you
offer?
I mean probably a wide rangefrom basics to more technical
application oriented subjects.
Speaker 3 (02:49):
Yes, uh at present we
have four short courses, so
they're certificate courses, andthen we also have an advanced
diploma in illuminationengineering, which is our
premier course, and that's atwo-year course.
The premier course covers anintroductory phase of lighting,
(03:11):
so getting familiar with theterminologies and things like
that, right through to learningabout the fourth industrial
revolution technologies, so it'sfully up-to-date.
All about lighting controls,interior and exterior design
techniques.
We cover over illumination indepth, light pollution in vast
(03:32):
detail as well, and then, ofcourse, lighting design software
is included in that package.
So, yeah, all the courses areon the learning platform.
All the courses are on thelearning platform, so our
students are able to have theirown student user account and,
depending on what course theyapply for, then they have access
(03:53):
to those learning resources andmaterials.
Speaker 1 (03:56):
And they get a
certification at the end.
Speaker 2 (03:59):
Yes, of course.
Certificates for the shortcourses and the advanced diploma
for the two-year course.
Certificates for the shortcourses and the advanced diploma
for the two-year course.
Speaker 3 (04:08):
Yes, our course used
to be accredited with SACA,
which is the South AfricanQualifications Association, but
now because we've grown quiteconsiderably on an international
front, so we are waiting forour accreditation with the
(04:30):
European Agency for HigherAccreditation and that's going
to mean that our course isinternationally recognized and
then students will be able toearn points for each of the
modules, which they could thenuse towards applying for other
courses around the world.
Speaker 2 (04:47):
Yeah it's like 120
credits.
Speaker 1 (04:50):
Oh, wow, okay.
So you get students from allover the world, don't you?
Speaker 3 (04:55):
Yes, we got a vast
footprint many students in the
Middle East, so the UAE, Qatar,Kuwait, Oman, Saudi Arabia,
Egypt.
Of course, many from SouthAfrica, not so many from Europe.
Maybe that's because of ourcourses are in English so it's
(05:16):
maybe a bit more difficult forpeople, maybe having to
translate and such.
But certainly where English isused as business language, I'd
say that's where our footprintis the biggest.
Speaker 1 (05:30):
Okay, so we've got to
talk about….
Speaker 2 (05:33):
Sorry, go ahead,
daniel, we've had as far afield
as Uruguay, argentina, hawaii,and in the US, of course, all
over and in Australia, yeah,yeah yeah, now we're going to
talk about AI and the impact ofAI on education.
Speaker 1 (05:52):
So maybe for
everybody to understand, how do
your students participate andhow do they submit their work
and what sort of things do theysubmit I assume assignments and
things like that so that we havethe whole picture on what we're
talking about and wherepotentially, ai may have an
influence.
Speaker 2 (06:12):
All right.
The students submit assignmentsonline, so we can have every
student working simultaneouslyon this platform and it's
available 24-7, 365.
On this platform and it'savailable 24-7, 365.
So it's a marvelous platformfrom an educational point of
view.
So they do it at their own paceor it's live.
(06:41):
Yeah, they work at their ownpace, but there's obviously a
study guide and we try and keepthem to the study guide.
A study guide and we try andkeep them to the study guide.
They determine their own studyplan based on the study guide
because everybody'scircumstances are different, so
you can't be constipate aparticular study go.
Speaker 3 (07:02):
Can I add also that
most of our students are
professionals, so architects,electrical engineers, interior
designers and, of course,lighting professionals.
So we don't have really schoolleavers, let's say, or high
school diplomas, arriving to ourschool.
It's more tailored.
The courses are for forprofessionals wanting to
(07:24):
increase their knowledge aboutthe lighting industry so what
would you say is the balance atthe moment?
Speaker 1 (07:30):
do you have more
professionals like light,
potential lighting designers,architects and designers, or are
there real, just students thathave just graduated and wanted
to learn the the discipline oflighting design before they
start a career?
Speaker 3 (07:45):
I think I would say
the majority are professionals.
So generally already working inthe lighting industry, either
maybe as sales persons, and nowthey've been thrust into a
situation where they need tolearn about design, for example.
Or then you have the architectswho are need to learn about
design, for example.
Or then you have the architectswho are wanting to learn more
(08:08):
about their unidentical twin,you know, because without light,
there cannot be anyarchitecture to be seen, you
know.
Speaker 1 (08:17):
Right, right, right,
right.
Okay Now, before I dive intowhat AI has as an impact in
education, I want to ask anindividual question to each of
you, because it will beinteresting to when both of you
are aware of AI.
But I would like to know howyou use AI at the moment, if at
all, in your daily life, what isyour knowledge of AI and if you
(08:41):
use it, potentially alsoprofessionally, in your daily
work.
It would be interesting to knowwhere you stand individually in
relation to AI.
Speaker 2 (09:00):
Sure, I think, Phil,
particularly in two different
ways.
The one is when I'm doingresearch on particular subjects
not always for students or forthe courses, because as you go
through life there are alwaysother questions that arise and
(09:22):
you can get an answer fromGoogle.
And you can get an answer fromGoogle, but you can get a far
more extensive answer from an AI.
And because we have a highlicense level with Microsoft, we
(09:42):
use Copilot, and I use Copilotquite a lot because it brings up
references.
Even so, it will actually giveyou references so you can go
back and you can seek moredetail from that specific
reference piece.
Speaker 3 (10:02):
And I'll also add
that, because we use CodePilot
as a team, that of course we useit for the school as well as in
the lighting design business.
So our AI companion, let's say,is trained on topics that we've
been prompting, you know, soit's developed its own, let's
(10:24):
say, writing style and thingslike that as well.
Speaker 2 (10:28):
Because with Copilot,
it actually searches our own
documents.
It will go and search all of myown documents on my local drive
or on OneDrive, being Microsoftusers, and it'll look for
answers that I've probably comeup with, or it'll pull all the
(10:51):
resources to give me answers,but then it will also go and
look outside.
Speaker 1 (10:58):
Yeah, yeah, yeah, I
use chat GPT a lot, which I feel
is also looking into what Ialready have, but obviously you
build up a library of thingswhen you work with it, and I use
Gemini and Copilot once in awhile, just sometimes to have
different opinions.
(11:18):
That's also good.
You can get one tool butsometimes two.
You can get one tool butsometimes two and have you used.
But maybe we come to that lateron because I'm also interested
to, if you're already into AI,whether you have used that to
check submission from yourstudents.
Speaker 3 (11:38):
So maybe on the
student side of things we'll
cover that next.
But in my daily life, um, Ikind of do the role of maybe
three or four people, um,because in our, in our business,
it's only phil and myself.
So in regards to our, ourstudents, I cover the the first
(11:59):
years and and phil moderates allthe second year's work.
So we've got that to cover.
But then in the lighting designbusiness I use AI quite
extensively.
I use Midjourney particularly soI have a license with them and
of course creating AI, let's saygenerating AI images for
(12:25):
concepts, is really great.
Another thing that I use it foris I work in Relux Lighting
Design Software to create all myreporting for projects and such
things.
But of course then it alsocreates high-resolution
ray-traced renders so I can takemy renders and put them into
(12:47):
Midjourney and ask Midjourney toanimate and create walkthrough
videos of my lighting designs.
So of course that's now anotherlike for me.
When I discovered that, it waslike kind of mind blowing
because the offering now that Ican provide to my clients is
like you know, this super wowfactor Instead of seeing this
(13:09):
flat, kind of you know renderfrom a project.
So it's kind of made things,let's say, more immersive in
that sense.
Speaker 1 (13:18):
Do you use the
language models maybe also to
structure your courses, or isthat purely the old-fashioned
way?
I'm pretty sure that Phil wouldhave?
Is that purely theold-fashioned way?
I'm pretty sure that Phil wouldhave done that in the
old-fashioned way writing downthe courses.
Speaker 2 (13:31):
No, it's the
old-fashioned way.
Yeah, yeah, I didn't use my AI.
Speaker 1 (13:35):
Yes, but I'm a bit
like you, phil, but I've now
discovered that these tools areso handy in structuring and
restructuring things likewhatever written document.
That could have a big impact onhow you further develop courses
in the future.
Speaker 3 (13:54):
So one thing we've
been researching.
Obviously, the possibilitiesare kind of endless at this
stage with AI, so onepossibility we're looking into
is taking our learning materialsfor each of the modules and
then you can create studentlessons and use your own voice,
(14:15):
to embed your voice into a videoor an animation to explain a
particular topic.
So that's a new route thatwe're navigating at the moment,
um, but it's like super excitingand I think the students will
benefit greatly from it.
So, instead of reading alearning module, um, they'll be
(14:38):
more engaged, you know, byseeing, hearing, you know,
absorbing with the eyes andtaking that well, really
absorbing that information intheir minds and not just reading
it from a transcript.
Speaker 1 (14:54):
Now let's get to the
key question.
Speaker 2 (14:58):
Another person says I
use AI, that's in accounting
software.
Speaker 1 (15:03):
Oh for sure, yeah,
there's.
Speaker 2 (15:04):
AI.
So I ask it at a point in time,at the end of the month or
early just after the month end,which students haven't paid
installments for those who payby installment, and it will tell
me who hasn't paid and thenI've prepared an email and I
(15:26):
just tell it.
Send them the email.
Speaker 1 (15:29):
Yeah, yeah, yeah, and
it sends the email.
Speaker 2 (15:32):
So I don't have to
sit and actually physically do
it.
So it's looking to the ways andmeans to use AI as another
employee.
Very good as another employee,which would do sometimes I have
to send a letter or somethinglike that to someone and I tell
(15:55):
AI, which is co-pilot in thisinstance, to actually prepare it
.
Give it good instructions andit will do the job.
Give it lousy instructions, itwill be a lousy job.
It will really be unacceptable.
It goes around what kind ofinstruction you give, just as
(16:17):
you would give a human aninstruction.
Speaker 1 (16:21):
The clearer the
instruction, the better the
outcome Correct.
I mean that's in general.
Yep, the clearer theconstruction, the better the
outcome Correct.
Absolutely, I mean that's ingeneral.
I think bigger companies arestarting to use that also to
track of their clients, and thepaymasters are making sure
everybody pays in time, and thenyou can train your AI agent to
look after that andautomatically send the messaging
(16:44):
when it's necessary.
Listen, I wanted to come to thekey question of why we're here
today, which is is AI cheatingor is it an actual evolution in
the whole learning process?
And that basically because atone point of time, you did start
to suspect that your studentswere using AI in their
(17:05):
submissions.
So the first how did you noticeit, how prevalent was it
already and what are yourthoughts on that?
What are we going to do now?
Speaker 3 (17:20):
Let's say in the
first year of our two-year
course.
Let's speak about that becausethat's where AI is incredibly
prevalent at this stage.
So the first year of ourpremier course is, let's say,
the majority of it istheory-based, so it's learning
(17:41):
about things like production oflight.
You know light sources.
So the assignments wereparticularly, let's say, you
give a question and the studentcan find the answers in the
learning module and then kind ofparrot answer it.
Let's say that in the past.
So of course we've been doingthis for a good few years and of
(18:03):
course we've been doing thisfor a good few years.
So from our history ofmoderating assignments, we could
easily see that as soon as AIwas introduced, particularly
just by the format and writingstyle of AI, the type of bullet
points, the type of heading soit's a very structured, let's
(18:23):
say, answer that AI gives.
It's really great, let's notsay it's bad, it's really great,
but it's incredibly easy toidentify.
And then of course the AI isnot resourcing any of the
learning material, maybe fromour learning models.
(18:45):
So when students provide ananswer for X question, then
maybe the AI isn't drawing onthe resources from our learning
module and then it doesn'tprovide the full answer.
So it's quite easy to see thedifference between pre-AI and
(19:07):
current situation with AI.
Speaker 1 (19:10):
So this started about
a year ago.
Speaker 2 (19:14):
Yeah, I'd say so.
It was during last year, duringlast year.
Speaker 3 (19:19):
But I'd say, in the
last three months particularly,
it has become like every justabout everybody recently has
used it.
Of course, then I know youreceived our recent newsletter
about ethical use of AI, so thisis how we got to today's
(19:42):
discussion.
So yeah, obviously we have toevolve with the times and what
we are looking at doing is akind of restructuring of the
course in terms of theassignments, and maybe we need
to look at making studentscreate answers in a more
(20:05):
creative manner or somethinglike that, let's say, where it's
not parrot fashion kind ofanswer, so it's more testing
their critical thinking ratherthan just providing an answer,
you know.
Speaker 2 (20:18):
Yeah, what we also
noticed with them answering and
using AI.
You could immediately seebecause that personal style and
perhaps the personal stylebecause they didn't have such a
masterful language ability toexpress themselves and all of a
(20:41):
sudden you can see like aperfect style of voice in style,
talking about the style, andimmediately you can see this is
not the person I know but thenice thing is you have some
comparison material before andafter so that you know the
content and the course materialitself.
Speaker 1 (21:03):
So I can imagine that
AI has the ability to dive, on
your side, into some of yourmaterials.
So that's pretty interesting.
You did mention, I think, thatyou were taking some action or
pausing maybe the courses tosort of reflect on the situation
before moving forward again.
(21:24):
Did I understand that correctly?
Speaker 2 (21:27):
No, we haven't paused
the courses.
What we're doing, I'm busy witha major refresh of the course.
It's a very significant refreshOn the basic theoretical side.
There's not too much you canrefresh, except in certain
(21:47):
instances in the newertechnologies.
But the typical old theory, thetheory, still applies.
Speaker 3 (21:56):
Yeah, when it comes
to lighting design techniques,
of course techniques likeRichard Kelly's are taught, you
know.
So the basis of design elements, that still remains the same.
Of course, there's newtechniques evolving with the
flexibility of LEDs, especiallyin the lighting control sphere
(22:18):
of things, and the amazingthings you can do with lights
nowadays.
Speaker 2 (22:23):
And all the new
technologies that we use in
lighting.
So you know there's a lot ofscope for it and I think it's
really good there is a place forAI there certainly is.
Speaker 1 (22:39):
Whether you want it
or not.
Speaker 2 (22:41):
Education, though, is
to determine what that place is
, so that students moveethically through the course.
Speaker 3 (22:51):
Yeah, and then you
asked about kind of what we're
doing about the AI use.
So we've, in the present moment, before we revamp, let's say,
the assignment papers, so we'veapplied a 15% penalty to the
(23:14):
total moderated grade for aparticular assignment.
Speaker 1 (23:19):
Explain the penalty.
How does that apply?
Speaker 3 (23:22):
So each of the
assignments are out of a certain
grade.
There's certain points.
Let's say x assignment orsomething?
yes, for if, if ai has been usedand it's, uh, been detected and
answers haven't been providedin full, you know, and obviously
we have that comparison to goby and when, when answers aren't
(23:44):
complete, or sometimes studentsask to provide a certain
diagram or create a presentation, this was actually a really
good example.
One particular question was tocreate a presentation.
So what the AI did?
The AI then created notes andsaid on slide one, include XXX
(24:11):
point On slide two.
And then the student copied andpasted this instead of
answering the question bycreating their own presentation,
which was the task.
So things like that, that'swhen the penalty is going to
apply well, I think people canargue.
Speaker 1 (24:29):
Sorry, go ahead, phil
.
Speaker 2 (24:31):
I had one where the
student answered certain
questions and actually only gotless than 50% of the points
because AI didn't source ourmodules with our manuals.
So just as an example, so thestudent actually lost out by
(24:56):
using AI and it's not AI.
Speaker 1 (25:01):
They can be addicted,
right.
So they sort of only use whatAI provides them and they don't
think further.
Speaker 3 (25:08):
Yeah, and then maybe
let's say, because the students
are learning at their own pace,so some students, for example,
have fallen behind with theirtime.
So now what you find is okay, Ineed to catch up, so I'm going
to use ai to answer all myassignments, you know.
So it's.
It's a really tricky place tonavigate this world of ai and
(25:33):
how it's.
It's kind of becoming a littlebit intrusive in the learning
sphere of things, in the designside of things.
It's incredible, and if youdon't use the tools that are
available now and learn them now, then you're going to be left
behind.
So you need to have theseskills and acquire these skills
(25:55):
now, before it's too late, andyou're left behind.
It's kind of like the computerrevolution of the 90s, you know.
Speaker 1 (26:05):
Yeah, yeah, but the
question is, how do we address
this?
Because people can argue thatAI is basically no different
from a calculator that you usein mathematics or these kind of
tools.
It has become a normal thing todo, so we need to find a way to
(26:26):
become transparent and somehowfind a way where we can draw the
line between what is your ownhonest contribution and what is
the assistance that you drewfrom the AI tool.
So declaring that you have usedit, or maybe even forcing
(26:48):
students to mention what theyhave used as a tool and how they
have used it, and maybe evenmentioning the sources, could be
a way, I imagine, to sort ofget a hold on this.
What are your thoughts on that?
Speaker 2 (27:04):
I think it is a
possibility to get a declaration
.
We'll include a declarationform.
It's something that we'veconsidered.
Excuse me a moment, but we hada student session last night.
We do it once a month, on thefirst Wednesday of every month,
(27:28):
where we have a student supportsession.
It's the opportunity wherestudents all come online and
they're able to ask anyquestions that they may have.
That is bothering them, but wealso took the opportunity last
evening to talk about AI and theethical use of AI and at the
(27:53):
same time, we also said AI hasgot a very important place in
the learning process, where youcan use it just as you've used
Google all of these years.
Ai is taking its place.
Google is falling short on itsanswers, whereas AI is far more
(28:17):
comprehensive in its answers.
It might not cover all thepoints to pass an assignment,
but it's very useful.
So at the end of the day, onceyou've got all this information
together, it's up to the studentto write the assignment and to
ensure that it addresses all thepoints to get full marks or
(28:43):
close to full marks.
Speaker 1 (28:45):
Then there's also the
human intervention in terms of
we all know that AI is notperfect and it sometimes
hallucinates a little bit andcomes up with things that are
potentially not correct.
So, encouraging students toreally double check what the AI
is providing them and mayberewarding, restructuring or
(29:09):
whatever it is that they need todo to make it their own and
make sure that what they pass onis actually correct.
Otherwise, you know, you cansay you can't say oh, the AI
gave me, this is not my fault.
In the end, it's still thehuman that makes the final
decisions and does the finalactions.
Speaker 2 (29:32):
I would like to give
an example.
The other evening we were at arestaurant and I was asked the
question by my dear wife why dothe lights of the city across
the bay twinkle?
And I gave an answer.
I'm not going to spend time onthat now, but interestingly,
(29:55):
when I got home I checked withAI to see what answer it would
give me.
With AI, see what answer wouldgive me.
Speaker 1 (30:10):
And I'm glad to say
that my answer was more
comprehensive than AI.
Speaker 3 (30:15):
Well, that's your 40
or 50 years of experience, bill,
yeah, totally.
But I think also one of the bigthings in the learning process
is considering the cognitiverisks in using AI for academic
tasks.
So there was recently a studyat MIT Media Lab and Cornell.
So they discovered thatstudents using AI for essay
(30:40):
writing showed up to 55% lessbrain activity, and these were
in areas like memory, creativityand semantic processing.
So many couldn't recall orquote their own essays and fell
quite short on their submissions.
So AI's reliance may hinderdevelopment of critical academic
(31:04):
skills like thinking, memoryretention and creativity.
Speaker 1 (31:10):
Well, I have to say
you may be aware of it, I'm
currently trying to develop acourse for lighting designers in
using AI.
Quite honestly, I don't knowmuch about AI, but the person
I'm working with is an AI freak.
Speaker 3 (31:30):
Yes, I've seen his
work.
Of course, I follow him onsocial media.
You've got to keep up with thegame.
Yes, yes, that's absolutelymind-blowing the possibilities.
Speaker 1 (31:42):
Yes, yes, yes and
that's why that's absolutely
mind-blowing possibilities yeahyeah, yeah, but that's why I I
walk this with him because I'mlike the human consciousness to
make sure that what he comes upwith is actually doable.
Yeah, um no, if you've been inthe industry, like me, for so
long time, a lot of things is isgut feeling and intuition.
We know when we see something,we know instantly whether it's
(32:03):
right or not.
We don't always have theexplanation, but we know, we can
feel it.
Speaker 3 (32:09):
Yeah, you can't fool
the eye.
Let's say that.
Speaker 1 (32:12):
Absolutely and that's
the relationship we have in
building up this course is hecomes up with AI stuff and I'm
saying, well, listen, this ishow we do it, and I'm trying to
sort of counterbalance themachine learning of the AI with
my human intuition.
So it's quite an interestingprocess and I'm learning along
the way, of course, as well.
(32:32):
But yeah, so now, having saidall that, what are you planning
in regards to implement AIpolicies, making sure that you
can define what is acceptable,what is unethical, when you get
them to do their coursework andexams and things like that?
(32:54):
Have you already managed tosort of define and finalize a
bit your views and how you wouldlike it to work?
Speaker 2 (33:03):
I'd like to first
answer some of it.
I've been following it veryclosely to see what some of the
great universities that we haveare doing about AI and just
recently, cape Town University,which is probably regarded as
(33:23):
one of the top universities inSouth Africa, it's actually
announced that they are going toallow AI to be used.
So they've gone the completedifferent way to allow them to
use AI, to allow them to use AI.
I'm not sure that I agree withit, because I can see what has
(33:48):
been happening and I'm sure,daniel, you've probably had a
similar experience where they'veused AI to answer, but they
don't get all the points thatthey would normally have got
when they did the assignment andimmediately you can see the
student is actually.
(34:09):
It's almost like aretrogressive step using AI,
whereas AI should be aprogressive step.
Interesting.
Speaker 3 (34:27):
Yeah, I kind of agree
, especially in the learning
sphere.
So, as I mentioned a while ago.
So it's just about that memoryretention.
Ai is not going to remember foryou is what I believe.
It's there as a companion, likeyour calculator, to assist you.
If you don't go on thislearning journey yourself, then
(34:53):
you are going to feel the sideeffects of using AI.
Someone's going to ask you aquestion and you're not going to
probably know the answerbecause you can't remember.
It is my opinion.
Speaker 1 (35:07):
You have to ask AI
first before you can answer?
Speaker 3 (35:09):
Yes, precisely.
Just hang on a minute.
Let me just send a voice noteto my AI to provide me an answer
and I'll get back to you.
Can we stop the conversation?
I think the learning journey isvery important, that you use AI
as a companion and not as areplacement, and then the best
(35:31):
way to learn, as I probably saidearlier as well, is the seeing,
the writing, the listening, thewatching you know, so that your
brain is absorbing all of thatinformation and then you're able
to have open discussions withyour colleagues and your network
or your clients, and you can bein the industry with confidence
(35:53):
.
Speaker 1 (35:54):
Yeah Well, I think
you can also refer back to the
older days when we used to dohand calculations.
I can still calculate prettywell by hand.
I don't need a calculator to dothings I can hand sketch.
I don't need a computer programto sketch something out right.
That's what you have trained inthe older days.
(36:16):
But at the same time I know howto use a calculator.
I know how well I'm not verygood in software to do renders,
but you know it's evolution,obviously.
And we need to somehow educatethe students on what the
limitations and the risk will beby using AI without human
(36:37):
oversight.
I think that will be one ofmaybe your task as well as
educators to try and bring thatbalance Like listen.
You need to train this part aswell, not only train the core,
but also train to think.
Speaker 2 (36:52):
You see Martin, if we
look back at how the course is
structured.
When I developed the course, Ideveloped it on the building
block principle.
I developed the course.
I developed it on the buildingblock principle.
What you learn in Module 1,you're still going to be using
it in Module 25.
Yes, it's not something that youlearn it and discard it.
(37:15):
You take it with you every stepof the way because that's
growing your lighting journey.
It's growing the expertise thatyou need in this world today to
do lighting successfully.
So you need to take all of itand not some of it.
And if you're going to use AI,there's some of it you're not
(37:38):
going to take because the bitthat you're relying on is going
to have shortcomings.
Yeah, so that's also a veryimportant fact from the way
we've structured the course.
Speaker 3 (37:55):
Yeah, and we're quite
open about it with our students
.
So, like last night, wediscussed it with them and they
gave their opinion.
So, like last night, wediscussed it with them and they
gave their opinion, and I thinkthe general consensus is that it
should be used as a companion,but not to replace.
So it's about spreading thatmessage and the risks that are
(38:15):
involved, especially in yourlearning journey.
But, of course, at the sametime, we're encouraging the use
of AI, because it's definitelyan assistant.
Speaker 2 (38:25):
And in the new
content which will come on
stream at the beginning of nextyear, from 1 January, it will
include some new content on AI,because AI is working into the
lighting sphere as well, so itwill be very important that they
(38:48):
keep up with it.
There's a question I want toask you, please, martin.
You were speaking about usingChatGPD.
Are you using the paid versionor the free version?
Speaker 1 (39:01):
Yeah, I've got the
pro version.
Speaker 2 (39:02):
You've got the pro
version.
You've got the pro version.
Now there's a very If peopleuse chat GPT, there's a very
different answer between thefree version and the pro version
.
So it's very significant.
So you get what you pay for.
(39:23):
It's very, very hard, so youget what you pay for.
Speaker 3 (39:25):
It's very, very hard,
of course.
Yeah, and then Gemini has adifferent answer, because it's
sourcing from websites allaround.
Yes, from all the differentaddresses.
So it's a very different answer.
That's provided becausesourcing from Google-based webs.
Speaker 1 (39:42):
Yeah, well, there's
also the issue of privacy and
confidentiality.
Obviously, you have enterpriseversions where it's really
contained, where whatever you dostays supposedly within your
own boundaries, right,everything is free, obviously is
accessible to everybody.
Everything is free, obviouslyis accessible to everybody.
(40:02):
So also for your coursematerials, which I would imagine
, even though it's based oncommon knowledge, is somewhat
proprietary to you, to yourschool, obviously you would like
to keep that also withincontrol and be sure that AI does
(40:25):
not start sharing around allyour proprietary work that you
have built up over the years.
Speaker 3 (40:34):
Yeah, it's a tricky
one to navigate, but of course
we are doing our best to keep upour work copyrighted and away
from AI.
So yeah, because our learningplatform is private with Moodle,
the outside AIs can't accessthat information.
Speaker 1 (41:00):
Yeah, I have a
question to both of you, because
there's obviously agenerational difference between
you, phil, and your son, and Ihave the same thing with my son.
Like we were brought up in theolder I mean I'm talking as the
older generation in the yearswhen I started and I feel the
(41:21):
same for you we didn't havecomputers, we had to do
everything by hand and then thecomputers came in bit by bit and
all that, and so we are broughtup differently.
But today's generation isgrowing up with all these tools
at their disposal.
So I could imagine that yourthoughts, phil, about what the
(41:44):
evolution of technology means isa bit different than Daniel
thinks of it.
Do you have any clashes betweenyou or different opinions?
I would imagine there is, butmaybe not.
Speaker 2 (41:56):
I don't think we've
had any clashes.
Speaker 3 (41:57):
I don't know.
We haven't had any clashesbecause, I must say, phil is,
like, incredibly knowledgeableabout technology.
He likes to keep up to date,and likewise with myself, and
we're constantly sharingdiscoveries with each other and
what's best for our business,how can we improve, how can we
(42:20):
make workflows faster, thingslike that.
So, yeah, I don't think we haveany clashes and I'm not that
young I'm closing in on the 4-0nowadays.
So yeah, although we'regenerations apart, we're quite
(42:42):
close to thinking the same.
Speaker 2 (42:44):
Yeah.
Speaker 3 (42:45):
Yeah.
Speaker 2 (42:46):
No, that's absolutely
true.
My philosophy is I've got tostay ahead, almost ahead of the
technologies, if you understandwhat I'm saying.
I've got to stay right up there.
It's my business to stay upthere, because how can we offer
(43:07):
our students right up-to-dateteaching if I'm staying there?
So it's really a major missionand it in fact keeps me young.
Speaker 1 (43:23):
Well, I think I'm
pretty much like you, phil,
because you have to be curious,you have to embrace new
technology.
We already had that over theyears, even in the lamp
technology, the lightingtechnology that is evolved from
simple fluorescent metal halidestill where we are today with
sophisticated smart LED systems.
We had to keep up and I thinkbeing curious keeps you young as
(43:47):
well.
And, yeah, I'm the same.
That's why I've embarked at myage to still do an AI course and
you're thinking why Just keepit easy?
But it's a way to keep up withit and be able to remain
relevant in this world, I think.
But we have the advantage thatwe carry that years and years of
(44:10):
experience and intuition withus, what we have learned from
the very early days of doingthings by hand and really
knowing what we're talking aboutby hand and really knowing what
we're talking about Whilst.
Today, if you develop a projectthat is mostly AI-driven, my
question would be well, whereare you going to gain that
(44:32):
experience?
Where is it that you're goingto go to that process purely by
yourself?
Learn from your mistakes notfrom the AI mistakes, but your
own mistakes right to build upyour expertise and your
experience.
I think that's a question thatI'm asking myself how the new
(44:52):
generation, today's generation,is going to build up that
experience if it's beingsupported and maybe even led by
AI.
Where's your part in all this?
Speaker 2 (45:05):
Yeah, I agree, 100%,
100%, because you know, if
there's a total reliance on that, on AI, there won't be
expertise.
No, so when you've got toaccount for what you've given
(45:26):
out there, who are you going tochallenge AI?
Speaker 3 (45:32):
and the.
AI in the lighting design spherein terms of projects, the AI
isn't going to be able to signthat certificate of conformity
at the end of the project.
When you have to sign it off,you have to do it and you have
to know exactly what this designneeds to achieve in various
applications.
Do I need 100 lux?
(45:53):
Do I need to light up thisemergency escape route correctly
?
How do I do it and how manylight fittings do I need on this
escape route, for example?
So, yeah, the AI is not goingto, firstly, in an academic
sense, remember for you and it'snot going to be able to sign
off your projects, but it cancertainly help, so definitely in
(46:13):
the the concept stages andproducing nice renders and great
ideas and things like that.
Um, because I do that for myprojects and just from a playing
around point of view, it's alsoreally great to see what can be
developed.
But, yeah, ai is not going todo your job and it doesn't have
(46:35):
your knowledge in a practicalsense.
Speaker 1 (46:38):
Yeah, no, absolutely.
And while AI will probably bevery good in theory, providing
your internet is working, butyou also have to build in the
emotions and the feelings.
Where are those?
The AI doesn't have that, atleast for now.
I would not be surprised thatthings like that will, of course
(47:00):
, be able to machine learnemotions and feelings, maybe as
well, but for now, the intuitionthat I mentioned before and the
emotions and the feelings thatwe have when we see lighting
lighting reacts differently todifferent materials, different
colors.
How well is AI able to pickthat up and transfer that into
(47:23):
emotional values?
That is something that is yetto be seen.
I think yeah.
And that's the part that I'msort of working on in the AI
course, because that's what Ifeel is missing.
Speaker 3 (47:37):
Yeah, maybe in the AI
sense in terms of creating
renders and things like that.
So let's maybe talk about apractical example of, maybe, a
bar counter and we want to washlight on the bar counter or down
from the bar counter andwashing the woodwork or
(47:57):
something like that.
Ai might be able to make areally nice concept and have
that bar glowing and lookingreally good, this nice warm
tones on the wood and thingslike that.
But then in a practical sense,how far should I mount that neon
(48:19):
flex, for example?
How far should it be from thebar counter?
What effect works best?
Should I create a lipunderneath the bar counter to
conceal something?
So in a practical sense, the AIdoesn't know what you need to
do in a lighting designtechnique sense, but from having
(48:39):
experience in a lighting designposition, you'll know what
works best in a practical sense.
Speaker 1 (48:49):
Yep, yep, yep,
absolutely.
So what do you think?
If you're going to integrate AIas part of the core materials,
I would imagine that at onepoint in time, you will have to
somehow manage, monitor andcontrol that aspect.
There are obviously severalsteps in the design process
(49:13):
where it could help, not only indeveloping narratives or
storytelling, but also in IDcreations and maybe making mood
boards or doing lightingcalculations.
I did one exercise with Sahilabout checking dialogues
(49:34):
calculations I'm not sure if youthought I can share that with
you where we asked AI to checkand fact check the lighting
calculations and show us wherewe could improve or what could
be, you know, correct.
So it's really interesting andit came out with some really
(49:55):
interesting things.
I can share the document withyou.
That would be amazing.
Speaker 3 (49:59):
Yeah, I'd love to
read through that.
Yeah, yeah, yeah.
And certainly, if you'recreating Latin specification
documents, you'll be able togenerate your report whether you
work in Relux or Dialux and getyour bill of quantities and
then ask the AI here's theproducts, here's the quantity
(50:21):
now gather all the information,insert the links of the product,
the images, et cetera, etcetera, and then generate this
lighting specification documentand of course, that saves an
incredible amount of timecreating a simple document.
You know.
Speaker 1 (50:38):
Yeah.
So there's obviously twodifferent approaches I think
Phil mentioned.
On one hand, you can do acontrolled integration of AI.
On the other hand, you can alsogive them a free hand and say
well, as long as you maybedeclare what you've been doing,
do whatever you do, do whateveryou want to do.
(51:00):
In the end, it's the end resultthat counts.
But if you look at it in thecontext of education, I think
it's a bit different because aslighting designers, I know many
lighting design practices thatalready use AI and put it as
part of their contract that theywill use AI in the development
(51:24):
of their design concepts.
So that is already sort of atransparency clause in the
contract where it said we willbe using AI.
But in education I think it's abit more tricky because on the
one hand, you want people tolearn and develop this right,
(51:46):
develop the brains and you knowcritical thinking, but on the
other hand, you also need togive them some sort of free hand
to use the tools that areavailable to work with.
Speaker 3 (51:59):
So it's a fine
balance, I think.
Well, in our two-year course,for example, we have the
end-of-year exams for first yearand second year.
At the end of first year theyhave to write a practical exam
and the practical exam consistsof Lumen method and point method
(52:21):
calculations.
So good luck to your AI.
Doing those calculations iswhat I'll say, because, yeah,
it's pretty tricky.
And that practical exam, yeah,it's giving students the
foundation of understanding whatlight does with mathematical
(52:42):
lighting calculations.
So yeah, I think, in the examsense, that one will be very
tricky to navigate with AI.
Speaker 2 (52:52):
And even the final
exams in second year.
I mean, both exams theory andthe practical are eight-hour
exams, Right, which is a wholeday if you think about it, and
there is no time to waste.
(53:12):
You can't afford to sit askingAI and tasking AI to get answers
and run the risk of not makingthe grade.
And I think that some of thelessons that students have
learned and I doubt this willcontinue to learn along the way
(53:34):
they are going to find that theyare not going to achieve the
results because the answer isnot a full answer or whatever
the case might be.
But will they know that?
You will see that, probably,but case might be, but will?
Speaker 1 (53:50):
they know that you
will see that probably, but they
might not see that.
Speaker 2 (53:54):
They will see it when
they get the results.
Yes, of course, without evenpenalizing them, ai is
penalizing them anyway.
Yeah, and maybe while we'retalking about, while's actually
a very interesting one.
Speaker 3 (54:10):
While we're talking
about penalizing, when we have
implemented the 15% penaltypolicy on certain assignments
that have been submitted, thestudents haven't contested a
single penalty.
Not a single person hascontested to us.
Speaker 1 (54:28):
So your penalty is
not high enough.
Speaker 3 (54:30):
Yeah, probably not.
Speaker 2 (54:33):
They get the penalty
and they don't even get the full
marks because AI falls short onthe answers.
So it's like a double whammy.
Speaker 1 (54:45):
I mean, I was always
told it doesn't matter how you
get it, as long as you get it.
So even if you don't have fullmarks, as much you can get it.
Speaker 2 (54:53):
I think AI is likened
to you know, you go to class,
you've got all your answerswritten on the palm of your hand
or on the back of your cup.
Speaker 1 (55:07):
But I really like the
idea of maybe not penalizing
but developing a point systemthat is with or without AI, so
maybe penalizing might not bethe right warning Maybe a reward
system is the better way to gowith using AI.
Speaker 3 (55:27):
Maybe that's the way
we could look at focusing it in
rather rewarding situations.
Speaker 2 (55:31):
That's an interesting
one.
Speaker 1 (55:36):
I like the idea it
might also promote people to get
extra points by not using AI orsomething like that.
Speaker 3 (55:45):
Many of our graduate
students have applied for
positions overseas.
So obviously our South Africanstudents and the situation
politically in the country rightnow is in a tricky place.
So many of the skilled workersare emigrating to other
(56:06):
countries around the world areemigrating to other countries
around the world and, as aresult, many of our graduate
students are applying forpositions at firms around the
planet.
So one of them is working atNolte Plus in London.
So we're very proud of thatachievement and he got the
position with the recognition ofthe school as one of the main
(56:31):
actors for being employed bythem.
So you know many at Iguzini,which is one of the world's
leading lighting suppliers.
So you know, like I think, ifyou're a graduate, you pass with
the assistance of AI.
You go work at Nelty, forexample, paul Nelty, I think his
(56:54):
name and he starts speaking toyou.
I think he's going to recognizethat you don't have all the
knowledge that you said you didif you work entirely with AI.
Speaker 2 (57:05):
Paul and Mark will
both see it straight away.
Speaker 1 (57:08):
You know, by the way,
nulti bought my company, right?
So I'm part of Nulti nowadays.
Yes, yes, reallyCongratulations on that.
But no, no, we are goingthrough that process now, also
within Nulti, to see what wouldbe the best way to integrate AI
in the whole design process.
(57:30):
It's not an easy one, but it'scertainly something, and that's
why these kind of discussionsare also helpful, which I also
share within the team, becauseit's important to understand
where we are at, what kind ofpractices are being applied in
different disciplines, howpeople use AI, and I think Paul
(57:53):
is totally right If yourknowledge is based on AI, then
maybe it's not really thatstrong, because you need to be
able to work also without acomputer at your fingertips.
So, yeah, it's a reallyfast-evolving platform at the
moment.
(58:13):
You mentioned that you use,like Midjourney and Gemini,
chatgpt, maybe a bit.
Are there any other tools thatyou're already looking at in
terms of, because there's alsovideo generation, not only the
text and language models, butwhat are you looking at at the
(58:35):
moment?
Speaker 3 (58:37):
So, from a design
sense, it's mid-journey
predominantly at the moment.
As I mentioned earlier, it'sbasically making video
generations of my design workout of Relux as my preferred
lighting design software, andthen, of course, then ChatGPT
(58:58):
have new models.
So, in terms of planning andproducing plan documents and,
say, circuitry documents,chatgpt and the new models that
have been released I think justthis week I saw one released
yesterday where you can uploadyour floor plans or your
(59:18):
elevations or section drawings,and ChatGPT will now create a
nice little model, 3darchitectural model, that like
spins kind of thing, and youcame out with a student model.
Yes, yes, yes, you shared thelink with me.
So of course, I was discussingPhil and I had a chat yesterday
(59:42):
just discussing the learningmodel.
Of course, with the learningmodel you can learn all you like
, but you're not going to getaccredited degree or diploma,
you know, yeah.
So I think if you're seekingeducation, then seek an
(01:00:04):
accredited educational program,and there's many to choose from
and many to to suit variousbudgets, you know.
So, yeah, ai is certainlysomething that we need to
embrace, but, yeah, as acompanion, as a companion and
not as a replacement okay.
Speaker 1 (01:00:27):
So, um, I think, uh,
we, we instrument Okay.
So I think we try.
I don't have, unless you have,still many other questions.
I'm sure that over the timecoming there will be more
questions as we develop thingsand we get hold and we get a
good understanding how AIimpacts the studies.
But right now, the studies.
(01:00:54):
But right now, where do youfeel you are at and how do you
want to continue with yourcourses?
You sort of gave already acouple of directives in terms of
what you're planning to do, butcan you sort of summarize that
where the BHA school is in termsof AI and how you plan to
embrace and integrate AI in thecourses?
Speaker 2 (01:01:14):
Yeah, I think the big
thing is we're going to embrace
AI as a subject, as a topic nota specific AI variant, but
particularly just as a topicvariant, but particularly just
as a topic to put it into focusand into perspective, from a
learning point of view, what thebenefits are and the pros and
(01:01:38):
cons of AI in the educationalenvironment.
You know, I understand, that theUniversity of Cape Town are
allowing free use of AI.
I think the decision has beentaken without a lot of thought,
but they also have thepopulation of the university
(01:02:02):
goes across the differentpopulation groups in the country
and the one population group.
Obviously the numbers are muchgreater and if you were to
penalize them, there would bemajor issues and possibly even
(01:02:24):
unrest and so on.
So the universities have a farmore difficult time of it than
we have.
So I think we have to get thatinto perspective as well.
Whereas we're a privateorganization, you choose to
study with us and that's it.
You have to abide by the rulesthat we set.
(01:02:48):
But we'll do it in such a way,I think, for the benefit of our
students, teach them what AI isand put AI in its place.
Speaker 3 (01:03:02):
Yeah, there's
definitely a place for it, of
course, and yeah, so in ourstudent support sessions, we
speak about, about ai, how wecan effectively use it as well.
I've shown the students someexamples of things that ai can
create, um, so, yeah, it'sthat's about embracing the
technology.
At the end of the day, as Isaid, you're going to be left
(01:03:24):
behind if you don't do it now.
So, coming up in in october, wehave for us is our, from a7 as
our, as our guest speaker forfor one of our upcoming events.
So, yeah, we are fully focused,we, we want to share the
knowledge, um, and, of course,share the the educational
(01:03:46):
experience with as many peopleas possible.
I think we believe it's ourrole to share knowledge, yeah,
and to give our students thegreatest educational experience
that they can afford.
Speaker 2 (01:04:01):
Yeah, and I think
there's another useful bit of
tech that we use.
It's Otter AI.
Have you heard of Otter AI?
Speaker 3 (01:04:14):
No, it's your
personal note taker.
Speaker 2 (01:04:18):
Okay, it's your
personal note taker.
So when you have a session withour students, our personal note
taker is there and it takesnotes for us, so you don't have
to bother yourself with makingnotes all the time.
Speaker 1 (01:04:32):
Yeah, oh, yeah, no
zoom has that as well.
Speaker 3 (01:04:35):
There's many
softwares that have like this ai
note taking uh, verbally, likeliterally every single word that
everybody says yeah and it doeslike if you, if you're showing
or doing screen shares, forexample, you're showing a
student last night wanted toknow what the difference is
between an asymmetrical lightdistribution pattern and a
(01:04:56):
batwing, so I could do thescreen share and then the
students could view what I'mtalking about.
But then when you review the AInotes, it's taken screenshots
and included that as part of thediscussion.
Speaker 1 (01:05:12):
Oh, okay, yeah, it's
so clever.
Speaker 3 (01:05:16):
Yes, so, like
students who didn't make it last
night to the support session,they'll receive the link a
little later to be able toreview the session notes.
Speaker 1 (01:05:26):
Yeah, yeah, yeah,
okay yeah.
I mean, that seems like asimple intervention for AI to do
.
Speaker 3 (01:05:31):
Yeah, yeah, Easy
peasy yeah yeah yeah.
Speaker 1 (01:05:34):
Listen, it has been a
great discussion.
I would like to finish up withgiving you each an opportunity
to maybe send out.
What kind of message would youhave to the students that are
going to use AI in theirlearning journey with you?
What would be your message tothem?
And yeah, I think that would beprobably a good parting shot to
(01:05:58):
end this discussion.
Speaker 3 (01:06:00):
Yeah.
Speaker 2 (01:06:01):
I would say you have
to take note of AI.
You have to start to get yourhead around AI.
Get your hands on it, startusing it.
Speaker 3 (01:06:12):
Whether you want to
continue using it it's a
personal decision, but you can'tbe left behind because,
technology moves at anincredible pace, so I would
encourage students, but I'dencourage them point out the
ethics of using AI as well timesthroughout our discussion, but
(01:06:51):
it's crucial to remember thatduring the learning journey,
there's no better way toreaffirm what you've learned by
reading, writing, listening anddiscussing these topics, because
AI is not going to remember itfor you, only you will.
Speaker 1 (01:07:03):
Guys, thank you so
much.
I would love to catch up, maybein half a year's time again, to
see where yeah, yes.
Speaker 3 (01:07:11):
I'll think of a
strong voice.
Of course, yes, totally no,it'll be great.
Thank you, that's been a realprivilege.
Thank you very much, martin.
Speaker 2 (01:07:18):
Thank you, martin,
and thank you, phil.
Speaker 3 (01:07:21):
Yeah, cheers guys,
thank you, thank you.
Speaker 2 (01:07:24):
Cheers.