November 25, 2025 • 23 mins
In this Reality Bytes reunion, Tom, Sean, Tim, Oriana and Megan unpack the buzzy rise of vibe coding — the AI-assisted development trend coined by Andrej Karpathy and already explored by companies like Meta and Microsoft. The panel digs beneath the hype: from accelerated prototyping and accessibility gains to serious risks around technical debt, shadow applications, governance, security and the loss of human accountability. Oriana and Megan highlight the importance of schema, context and genuine creativity, while Tim warns against mistaking speed for quality. Is vibe coding the future - or just another fragile shortcut? The panel isn’t convinced… yet.
Learn more about Spark, The World’s Most Powerful DEX Agent for Employees, hereÂ
Download The First Annual Workplace Productivity Report hereÂ
Learn more about Spark, The World’s Most Powerful DEX Agent for Employees, hereÂ
Download The First Annual Workplace Productivity Report hereÂ
Mark as Played
Transcript
Episode Transcript
Available transcripts are automatically generated. Complete accuracy is not guaranteed.
Speaker 1 (00:01):
Welcome it, change Makers to the Deck Show with Tim
Flower and Tom McGrath. Let's get into it.
Speaker 2 (00:09):
Hello, change Makers, it's reality bites.
Speaker 3 (00:11):
We're bringing the band back together.
Speaker 2 (00:13):
We've got Orianna, We've got Megan, We've got Sean, and
we've got Timbo. I say, we laugh a little bit
because I just said we've got Oriana and Oreanna has
just disappeared from our recording studio, and poor Orianna. We
were having trouble last time around as well, so hopefully.
Speaker 3 (00:30):
Orianna returns and it's all good.
Speaker 2 (00:32):
So look, one of the you know, I think I
probably present a podcast because somewhere in me when I
was a kid, I quite liked the idea of being
a DJ.
Speaker 3 (00:41):
I think it's probably true, and as such, when the
suggestion was.
Speaker 2 (00:45):
Put forward by our very own sure Maulvi, but we
maybe do an episode on vibe coding. I think it felt,
you know, quite dj ish as a terminology to use,
which is no reason to do a podcast in and
of itself. But then we did also have a fantastic
episode Tim and I last week where the conversational interface
(01:09):
was our focus walls. The conversational interface is kind of
just transforming the world as we know it right now.
So when we looked into vibe coding, we thought about it,
this might be an interesting topic to pursue. But outside
the name, Sean, outside the cool sounding term, what was
(01:30):
your motivation for covering this topic?
Speaker 3 (01:33):
Why did it catch your eye? Well?
Speaker 4 (01:36):
I think you and I Tom, we trend towards more
picking up vibes or vibes p vibe orientated, not so
much coding. So I think any any way to bridge
that gap.
Speaker 5 (01:49):
That was appealing to me? I like why I was
interested in the topic. And yeah, so let's back up
a second. So vibe coding coined by earlier this year,
coined by Andre Carpathy. He introduced the term for vibe coding.
It's AI assist its software development practice where developers use
natural language prompts to guide in AI in generating, refining,
(02:12):
and debugging code. And as Andre said, it fully giving
into the vibes, embracing exponentials, and forgetting that the code
even exists.
Speaker 6 (02:23):
And let me just note too, So this is more
than just kind of a fad.
Speaker 5 (02:28):
It is being used by companies like Meta, Google, IBM,
and Microsoft for some of the positives. People are saying
developers especially that it speeds up prototyping. So if you
want to throw away stuff, but you want to test
out something quickly, this can speed up that process. Obviously,
it opens up a lot of questions about if you're
(02:51):
skipping over line by line in code, are you opening
up the door for bigger problems later on? But you know,
as self confessed idiot with some of these things, I
find it to be very interesting and I would love
to vibe code one day.
Speaker 2 (03:09):
Weren't you vibe code with me? Yes, I say, Tim.
From an operational perspective, from an infrastructure perspective, does the
vibe code in concept ring any alarm bells?
Speaker 3 (03:20):
Fee?
Speaker 7 (03:22):
And so I started my educational career in software development,
So my traditional coding goes way back, even though I
didn't make a formal career out of it, and I
get to kind of play the downer on all of this.
Right the alarm bells. I did the same thing Sean.
(03:43):
I just to make sure Ike was in the right
frame of mind. I googled it and I came up
with vibe coding as a software development approach where a
user describes the desired functionality in natural language. Some time
goes back to our episode last week where we're using
language instead of a screen so much to generate things
and in this case, generate code. The one alarm that
(04:05):
the thing that started my alarm bells was the last
sentence of the description. This method accelerates the development of
prototypes and applications, making coding more accessible. So that's the
first thing that rang my alarm bell is is the
coder qualified to discern the output as viable and a
(04:27):
quality product. I tested it, and I said, build me
a website that does that, creates an ROI model based
on these metrics, and it came back with a website
that had some interesting visible things to plug in data.
But when I looked at the output and how it
really was doing the mechanics behind the scene, it wasn't viable.
(04:49):
But if you don't know, if you're just if you're
not an ROI person or a coder, you really if
companies are relying on that method to generate their applications,
you're at the mercy of what the AI model gives you.
So that was the first alarm bell for me. The
second one are you And maybe it's kind of a
(05:12):
linked to that concern, are you qualified enough as a coder.
If you've got to now take that as a template
and take it on. Do you have enough ability to
modify and improve the quality or are you just gonna
iteratively ask AI to make it better. No, that's not
good enough, make it better. So again you're at the
mercy of the AI model and the vibe coder to
(05:36):
get the product. The other alarm bells Tom for me,
were does it it's in this kind of realm of
making it more accessible? Is it really within their role?
We've talked about shadow it, We've talked about shadow AI.
Now this could open up a whole new world of
(05:58):
shadow applications. Is it really their role to develop an application?
Are you starting to cross boundaries into other areas. Can
a line of business say I can't get my IT
developers to create this application, I'm just gonna go do
it on our own. So, security, governance, all of that stuff,
application sprawl is a concern. And then the last one
(06:21):
is standardization. So in my previous organization at an insurance company,
we had five businesses and each of the businesses had
their own application developers, and some used Java, and some
used C plus bus, and some used modern application platforms
(06:43):
and nobody followed the same standard, and that risk here
is here as well. If you're gonna if this causes
a splinter, and now you've got applications developed with different vibes,
how do you handle integrations, how do you handle data sharing,
how do you handle application support? It really it doesn't
alleviate the need to apply standards and governances and controls
(07:05):
to alleviate all of this, and those things are solvable,
but it just making it more accessible to people opens
up a whole can of worms that you've got to
pay attention to.
Speaker 2 (07:16):
Really well, said Tim, And it was very interesting you alluded,
as I did, to the last episode that we released
with Rob Wilson, the Death of the screen.
Speaker 3 (07:26):
You have a death of the screen is they should
have called the episode of that.
Speaker 2 (07:30):
It was in fact called agentic AI at the end
of traditional it t and Rob Wilson is, among other things,
the fascinating author of the Age of Invisible Machines. And
it was a really eye opening conversation for me, and
it also connected very nicely with our chat with Pedro.
Tim and one things that the phrase Pedro used that
(07:53):
then Rob kind of explored in more depth was this
notion that AI differs from individual.
Speaker 3 (08:01):
Real employees, from real people in the AI.
Speaker 2 (08:04):
Has no skin in the game, and the workplace is
gonna continue to require people who take responsibility for outcomes,
people who take responsibility for things. And I just wonder
if that isn't really what you're flagging. Right obviously, you know,
the functionality here is phenomenal, But there's that is going
(08:27):
to require, as you say, standards is going to require
the human in the loop ain't going anywhere in this
particular part.
Speaker 3 (08:33):
The human in the loop.
Speaker 7 (08:34):
And I'll use an example. If you're in California on
San Francisco, which I just was recently, way More taxis
driverless taxis are everywhere. When a way more runs a
red light or gets an a fender bender, who gets
the ticket? Where's the human in the in the loop
to take accountability for that problem? Same thing with vibe coding,
right or AI.
Speaker 2 (08:55):
I say, blame the passenger, Timber, That's great, I'm onlying there,
you know, over passenger in jail.
Speaker 7 (09:02):
Without a doubt the passenger could have said, hey, run
this red light. You don't know what they said.
Speaker 3 (09:08):
Yeah, So it is interesting.
Speaker 2 (09:10):
You know, we definitely recommend people to check out that
conversation with Rob Wilson, Orianna fascinated who is who is
back with us ladies and gentlemen, I thankfully, uh fascinated
to get to get your insights here. In the chat
before the show, you did raise some concerns about the
(09:31):
question of technical debt and legacy systems and how I
guess this interacts as a challenge, as a standardization challenge
with his very very aggressive novelty encoding. What are your
what are your general reflections and positions?
Speaker 8 (09:49):
Yeah, so the thing that really stood out to me
when I started seeing this is how does it scale?
You know, because of all of the issues that Tim
has already re and because I really think about code
in the context of really really big, complex systems. Right,
(10:10):
I've only ever really worked in enterprise level software. It's
never really worked in like websites or apps or anything
where you can make this kind of fine tuning experimental
stuff without it risking really massive downstream effects. And every
single person in place I've worked with has had been
(10:30):
going issues of tech depth, challenges of improving legacy systems,
and also a lot of human problems that go along
with that, Right, So I always think about it in
terms of how do we make our systems human centered?
And I think that some people are going to see
vibe coding as a solution for that, right, Like, oh, well,
it's going to be human centered because I'm going to
(10:50):
customize it to what I want to do. But that
isn't really what makes technology human centered, right. What makes
technology human center is having a process that understands what
people actually need, not just what they think they need
in the moment, that interacts with different things and like
different components in a system in ways that are strategic
(11:15):
and thoughtful and can last right and can be built
on right. Like I think anybody who's really technical will
tell you that an alarming amount of the Internet is
built on very tiny, fragile components, where a very small
like a single issue in a single library, can have
(11:35):
cascading effects all over in an industry. We're recording this
right after having the cloud flare outage, so honestly, we're
feeling that very acutely right now. And I want to
connect that to something that Dina mentioned. I'm going to
quote Tina since she's not here to say it, but
she said that one of the problems with vibe coding
(11:57):
is that it can hurt dex because you're just getting
code from nowhere without schema or context, and I think
that scheme in context is so important to building really
robust code that you need, especially for anything complex.
Speaker 3 (12:12):
One hundred percent.
Speaker 2 (12:13):
What do you think do you think putting these kinds
of concerns not necessarily to the side, but shelving them
at least for a second. To what degree do you
think this is going to be able to answer the
demand or reflect that demand for more custom solutions by
employees themselves and just for a more nuanced, reflective, personalized system.
Speaker 8 (12:37):
Yeah. So I would say, have you ever used open
source software? And have you used good open source software
and bad open source software? So I listen to my
podcasts on something called Antennapod, which an open source platform.
It's really nice. It's super customizable. It's also a little
bit of a pain sometimes because you can customize everything, okay,
(13:01):
because it's an open source project and people wanted it
to do all the things. And one of the reasons
why I like it is also that it's quite lightweight.
So this is a passion project. People have put a
lot of effort into making it do these things in
a way that is effective, right, because it's a much
smaller app than many others and it has all these customizations.
(13:21):
But that requires a lot of skill and a lot
of effort, right, So that's sort of my positive example,
And even that is quite confoluted, right, And there's a
lot of open source projects where you can see that.
It's just it's a complete manner, it's buggy, it's got
lots of overhead. It you know, doesn't have basic features
(13:41):
that you really want, and then it has so much
customization that it's completely impossible to learn how to use.
And that's why I think that it's really important to
have a concrete point of view and a united vision
even when you're trying to think about how to customize
thanks to.
Speaker 2 (13:57):
People, any other things you can see where you would
anticipate there could be an impact upon decks, the ability
to effectively manage and monitor decks in an organization any ways,
it could jeopardize the use, for instance, of next being
products or any variety of sm tools that you can
(14:18):
think of.
Speaker 8 (14:21):
Well, when I think of how it could have a
negative interaction with DEX, I think part of that is
just you know, the dangers of shadowy Tea. The security stuff.
A lot of the stuff Tim talked about immediately keep
to my mind in terms of it's very hard to
monitor or manage anticipate issues, and especially some of you
(14:44):
who did previously work at a cybersecurity company, one of
the things that is very true cybersecurity is that it
is much much harder to defend because you're constantly trying
to find every single hole and you can never find
all of the possible holes. Of looking holes into your
systems at random at an alarming pace, that makes it
(15:08):
really really impossible to do security.
Speaker 3 (15:10):
Well.
Speaker 8 (15:11):
I do think the other side of dex application to
sort of the IT perspective, right, So I think that's
also a totally separate question in a way, is okay, well,
how can vibe coding help people who want to automate
processes who want to have custom solutions and technical resources
(15:34):
that they don't currently have, right? And I do you
think that's both a cool application, right? Like I do
think it's a possible positive venue for democratizing coding in
a way, but I do think that it needs systems
in place to manage it to avoid all of these
issues that we're highlighting.
Speaker 3 (15:56):
I love it we're moving inside out.
Speaker 2 (15:58):
I think everybody be styled with all the cautionary elements,
and we're moving towards the positive visionary usages.
Speaker 3 (16:06):
Megan, bring us all the way there.
Speaker 2 (16:08):
If you would put putting some of these flaws to
one side, what ways do you envisage or are you
hearing about from the customers you speak to so often
that the vibe coding is beginning to help individuals or
deed organizations.
Speaker 9 (16:23):
Wouldn't that be funny. I can't say that I've talked
to anyone in our customer base that is using vibe
coding at the enterprise level, and I think that that's
probably due to the cautionary points that we've made so far.
I think scalability right like what Oreana just mentioned. And yeah,
not to totally just not non answer your question, but
(16:47):
another aspect of this that I was thinking about, obviously,
this vibe coding thing. I'm not a coder at all,
so I wouldn't I don't think I would ever try
it because I have no way of knowing if what
the code it spits out as useful or not. I
think about my own use of these Genai tools, and
I use them right to do some to draft copy
for me to do writing to. Basically, that's kind of
(17:09):
the only thing I use them for is to write
stuff for me. And what I've learned from working with
them for the last however long, is you have to
be really specific and what you want. If you just say, hey,
write me an article about this thing, it's going to
spit you out the most generic, boring like unreadable exactly
with the word count you ask for article with all
(17:30):
the creativity stripped out of it. And you have to
be very want article that covers this thing with this
audience and this tone in voice, cover X, Y and Z.
It's even better if you can give it an outline
and then it will you do something.
Speaker 2 (17:41):
But still, I.
Speaker 9 (17:43):
Mean, there's a whole thing like justice for the M
dash right, like nobody uses m dashes anymore because it's
such a sign that you're writing was written by Genai
and not by a human being. I think human beings
bring a creativity that Jeni just doesn't replicate. And I
think we've talked about this before, and I'm not a coder,
so I can't say whether that's happening with the vibe
coding as well. If it's just sort of the dumbification
(18:03):
of code output, like the way that it's become the
dumbification of writing output. The things that Jenai writes, they're
just not that inspiring, they're not that fun to read.
They're useful in certain contexts, right, like a marketing copy,
like a landing page or an email copy or whatever.
It follows a rope, standard set up, and so it's
okay that it's not like the most thrilling and exciting
(18:25):
thing you've ever read in your life.
Speaker 3 (18:27):
But I don't know.
Speaker 9 (18:28):
I think the beauty of having a human solve something
is that we come up with a creative way to
solve a problem. And it's that's the thing that I
think our customers get excited about with decks is that
it requires creative thinking to solve problems that previously it
didn't think they could solve. And now they hear a
pobly like, actually, you know, because we have this dex
data and we have this solution, we can solve for
that vibe coding, I don't know, maybe someone can will
(18:53):
hear me say this and come send me a message
on LinkedIn with all the different ways that vibe coding
has enabled them to be incredibly creative, and I would
love to hear that, but to me, it seems just
like another example of like we're just dumbing down the
output that we're getting out of these things. And every
example I've heard of people using vibe coding has been
to make those sort of small individual projects. A website
(19:16):
that can generate a goofy image, an app that can
generate something kind of silly, or I've been reading articles
about criminals using it to hack into things and ransom people,
which obviously isn't what these companies want. So at the
enterprise level, I think we're going to bump into all
the things we've already identified in this conversation. And I
(19:37):
have not heard about any of our customers using vibe
coding for their own internal software.
Speaker 2 (19:44):
It can sound fun, though what you use for you
What did you use tim for your experiment with it?
Speaker 7 (19:50):
I don't remember which tool it was.
Speaker 3 (19:53):
I'd have to go back.
Speaker 7 (19:53):
It was it was a few months back, but it
was it was very similar to Meghan, what you were
talking about, where the result that I got was very simplistic,
It was not as robust as I needed it to be,
and I really felt like pursuing that avenue to get
something better was going to be a waste of my
time because it just wasn't going to get there that
(20:16):
I would be able to do it much better. And
that goes to the skill of the person asking the question.
Maybe if I wasn't an ROI kind of a person,
I would have pursued it for a lot longer, I.
Speaker 3 (20:26):
Think, Tom.
Speaker 7 (20:27):
It also plays into some of the things we've talked about,
does a I really make you more productive or are
you chasing something that feels more productive but in the
end you're actually wasting more time.
Speaker 2 (20:37):
Interesting, and I should add, you know, congratulations to vibe
coding for being the ost Dictionary Word.
Speaker 3 (20:43):
Of for year twenty twenty five.
Speaker 10 (20:45):
But I will allow Sewan to finish this conversation off
with a prediction Sewan as to whether vibe cozing as
a trend is about to come crashing down in the
new year or we've only seen the beginning of its rise.
Speaker 6 (21:03):
I'm gonna say we're only seeing the beginning only because Meta, Google, IBM,
and Microsoft are willing to put it into training modules
and see what it can come back with, and they've
done it before in the past, where there's been big surprises,
So if they're willing to tinker with it, let's not
rate it off just yet.
Speaker 2 (21:22):
You don't want to go up against their collective might
too strongly. There's some strong there, well, I think I
think there's certainly something there, and I think there's certainly
some some guardrails suggested by a panel today which we
would all do all do well to keep in mind
as we as we continue to experiment heartily with AI
(21:44):
and in.
Speaker 3 (21:44):
So many different areas.
Speaker 2 (21:47):
But look, we've just come out of our I would
think of our series of shows celebrating marking our two
hundredth episode as a not just as the Deck Show,
but as Reality Bites as well. We all contributed to
ever to reaching this incredible numerical milestone, and we're just
entering that that period of the year where things start
(22:09):
to wind down and we're going to hear you know,
people who have been listening to the show for years
know that this panel we come together, we hash out
the year it was twenty twenty five and our highlights
and our predictions the next year, and we also Tim
and I will also be talking to returning show favorite
Jeff Right to hear how his years, is that right,
(22:33):
might not even have known that.
Speaker 7 (22:37):
By now we can assume is the perennial year.
Speaker 3 (22:39):
Then yeah, exactly right, exactly right.
Speaker 2 (22:42):
So we hope everybody listening joins us to that, it
joins us to those episodes, and thanks everybody for coming
on to discuss vibe coding and I'll see you very soon.
Speaker 1 (22:56):
To make sure that you never miss an episode, subscribe
to the show and podcasts Spotify or your favorite podcast player,
and if you're listening on Apple podcasts, make sure to
leave a rating of the show. Just tap the number
of stars you think the podcast deserves.
Speaker 3 (23:10):
If you'd like to.
Speaker 1 (23:11):
Learn more about how next Think can help me improve
your digital employee experience, head over to nextthink dot com.
Thank you so much for listening. Until next time,
Welcome it, change Makers to the Deck Show with Tim
Flower and Tom McGrath. Let's get into it.
Speaker 2 (00:09):
Hello, change Makers, it's reality bites.
Speaker 3 (00:11):
We're bringing the band back together.
Speaker 2 (00:13):
We've got Orianna, We've got Megan, We've got Sean, and
we've got Timbo. I say, we laugh a little bit
because I just said we've got Oriana and Oreanna has
just disappeared from our recording studio, and poor Orianna. We
were having trouble last time around as well, so hopefully.
Speaker 3 (00:30):
Orianna returns and it's all good.
Speaker 2 (00:32):
So look, one of the you know, I think I
probably present a podcast because somewhere in me when I
was a kid, I quite liked the idea of being
a DJ.
Speaker 3 (00:41):
I think it's probably true, and as such, when the
suggestion was.
Speaker 2 (00:45):
Put forward by our very own sure Maulvi, but we
maybe do an episode on vibe coding. I think it felt,
you know, quite dj ish as a terminology to use,
which is no reason to do a podcast in and
of itself. But then we did also have a fantastic
episode Tim and I last week where the conversational interface
(01:09):
was our focus walls. The conversational interface is kind of
just transforming the world as we know it right now.
So when we looked into vibe coding, we thought about it,
this might be an interesting topic to pursue. But outside
the name, Sean, outside the cool sounding term, what was
(01:30):
your motivation for covering this topic?
Speaker 3 (01:33):
Why did it catch your eye? Well?
Speaker 4 (01:36):
I think you and I Tom, we trend towards more
picking up vibes or vibes p vibe orientated, not so
much coding. So I think any any way to bridge
that gap.
Speaker 5 (01:49):
That was appealing to me? I like why I was
interested in the topic. And yeah, so let's back up
a second. So vibe coding coined by earlier this year,
coined by Andre Carpathy. He introduced the term for vibe coding.
It's AI assist its software development practice where developers use
natural language prompts to guide in AI in generating, refining,
(02:12):
and debugging code. And as Andre said, it fully giving
into the vibes, embracing exponentials, and forgetting that the code
even exists.
Speaker 6 (02:23):
And let me just note too, So this is more
than just kind of a fad.
Speaker 5 (02:28):
It is being used by companies like Meta, Google, IBM,
and Microsoft for some of the positives. People are saying
developers especially that it speeds up prototyping. So if you
want to throw away stuff, but you want to test
out something quickly, this can speed up that process. Obviously,
it opens up a lot of questions about if you're
(02:51):
skipping over line by line in code, are you opening
up the door for bigger problems later on? But you know,
as self confessed idiot with some of these things, I
find it to be very interesting and I would love
to vibe code one day.
Speaker 2 (03:09):
Weren't you vibe code with me? Yes, I say, Tim.
From an operational perspective, from an infrastructure perspective, does the
vibe code in concept ring any alarm bells?
Speaker 3 (03:20):
Fee?
Speaker 7 (03:22):
And so I started my educational career in software development,
So my traditional coding goes way back, even though I
didn't make a formal career out of it, and I
get to kind of play the downer on all of this.
Right the alarm bells. I did the same thing Sean.
(03:43):
I just to make sure Ike was in the right
frame of mind. I googled it and I came up
with vibe coding as a software development approach where a
user describes the desired functionality in natural language. Some time
goes back to our episode last week where we're using
language instead of a screen so much to generate things
and in this case, generate code. The one alarm that
(04:05):
the thing that started my alarm bells was the last
sentence of the description. This method accelerates the development of
prototypes and applications, making coding more accessible. So that's the
first thing that rang my alarm bell is is the
coder qualified to discern the output as viable and a
(04:27):
quality product. I tested it, and I said, build me
a website that does that, creates an ROI model based
on these metrics, and it came back with a website
that had some interesting visible things to plug in data.
But when I looked at the output and how it
really was doing the mechanics behind the scene, it wasn't viable.
(04:49):
But if you don't know, if you're just if you're
not an ROI person or a coder, you really if
companies are relying on that method to generate their applications,
you're at the mercy of what the AI model gives you.
So that was the first alarm bell for me. The
second one are you And maybe it's kind of a
(05:12):
linked to that concern, are you qualified enough as a coder.
If you've got to now take that as a template
and take it on. Do you have enough ability to
modify and improve the quality or are you just gonna
iteratively ask AI to make it better. No, that's not
good enough, make it better. So again you're at the
mercy of the AI model and the vibe coder to
(05:36):
get the product. The other alarm bells Tom for me,
were does it it's in this kind of realm of
making it more accessible? Is it really within their role?
We've talked about shadow it, We've talked about shadow AI.
Now this could open up a whole new world of
(05:58):
shadow applications. Is it really their role to develop an application?
Are you starting to cross boundaries into other areas. Can
a line of business say I can't get my IT
developers to create this application, I'm just gonna go do
it on our own. So, security, governance, all of that stuff,
application sprawl is a concern. And then the last one
(06:21):
is standardization. So in my previous organization at an insurance company,
we had five businesses and each of the businesses had
their own application developers, and some used Java, and some
used C plus bus, and some used modern application platforms
(06:43):
and nobody followed the same standard, and that risk here
is here as well. If you're gonna if this causes
a splinter, and now you've got applications developed with different vibes,
how do you handle integrations, how do you handle data sharing,
how do you handle application support? It really it doesn't
alleviate the need to apply standards and governances and controls
(07:05):
to alleviate all of this, and those things are solvable,
but it just making it more accessible to people opens
up a whole can of worms that you've got to
pay attention to.
Speaker 2 (07:16):
Really well, said Tim, And it was very interesting you alluded,
as I did, to the last episode that we released
with Rob Wilson, the Death of the screen.
Speaker 3 (07:26):
You have a death of the screen is they should
have called the episode of that.
Speaker 2 (07:30):
It was in fact called agentic AI at the end
of traditional it t and Rob Wilson is, among other things,
the fascinating author of the Age of Invisible Machines. And
it was a really eye opening conversation for me, and
it also connected very nicely with our chat with Pedro.
Tim and one things that the phrase Pedro used that
(07:53):
then Rob kind of explored in more depth was this
notion that AI differs from individual.
Speaker 3 (08:01):
Real employees, from real people in the AI.
Speaker 2 (08:04):
Has no skin in the game, and the workplace is
gonna continue to require people who take responsibility for outcomes,
people who take responsibility for things. And I just wonder
if that isn't really what you're flagging. Right obviously, you know,
the functionality here is phenomenal, But there's that is going
(08:27):
to require, as you say, standards is going to require
the human in the loop ain't going anywhere in this
particular part.
Speaker 3 (08:33):
The human in the loop.
Speaker 7 (08:34):
And I'll use an example. If you're in California on
San Francisco, which I just was recently, way More taxis
driverless taxis are everywhere. When a way more runs a
red light or gets an a fender bender, who gets
the ticket? Where's the human in the in the loop
to take accountability for that problem? Same thing with vibe coding,
right or AI.
Speaker 2 (08:55):
I say, blame the passenger, Timber, That's great, I'm onlying there,
you know, over passenger in jail.
Speaker 7 (09:02):
Without a doubt the passenger could have said, hey, run
this red light. You don't know what they said.
Speaker 3 (09:08):
Yeah, So it is interesting.
Speaker 2 (09:10):
You know, we definitely recommend people to check out that
conversation with Rob Wilson, Orianna fascinated who is who is
back with us ladies and gentlemen, I thankfully, uh fascinated
to get to get your insights here. In the chat
before the show, you did raise some concerns about the
(09:31):
question of technical debt and legacy systems and how I
guess this interacts as a challenge, as a standardization challenge
with his very very aggressive novelty encoding. What are your
what are your general reflections and positions?
Speaker 8 (09:49):
Yeah, so the thing that really stood out to me
when I started seeing this is how does it scale?
You know, because of all of the issues that Tim
has already re and because I really think about code
in the context of really really big, complex systems. Right,
(10:10):
I've only ever really worked in enterprise level software. It's
never really worked in like websites or apps or anything
where you can make this kind of fine tuning experimental
stuff without it risking really massive downstream effects. And every
single person in place I've worked with has had been
(10:30):
going issues of tech depth, challenges of improving legacy systems,
and also a lot of human problems that go along
with that, Right, So I always think about it in
terms of how do we make our systems human centered?
And I think that some people are going to see
vibe coding as a solution for that, right, Like, oh, well,
it's going to be human centered because I'm going to
(10:50):
customize it to what I want to do. But that
isn't really what makes technology human centered, right. What makes
technology human center is having a process that understands what
people actually need, not just what they think they need
in the moment, that interacts with different things and like
different components in a system in ways that are strategic
(11:15):
and thoughtful and can last right and can be built
on right. Like I think anybody who's really technical will
tell you that an alarming amount of the Internet is
built on very tiny, fragile components, where a very small
like a single issue in a single library, can have
(11:35):
cascading effects all over in an industry. We're recording this
right after having the cloud flare outage, so honestly, we're
feeling that very acutely right now. And I want to
connect that to something that Dina mentioned. I'm going to
quote Tina since she's not here to say it, but
she said that one of the problems with vibe coding
(11:57):
is that it can hurt dex because you're just getting
code from nowhere without schema or context, and I think
that scheme in context is so important to building really
robust code that you need, especially for anything complex.
Speaker 3 (12:12):
One hundred percent.
Speaker 2 (12:13):
What do you think do you think putting these kinds
of concerns not necessarily to the side, but shelving them
at least for a second. To what degree do you
think this is going to be able to answer the
demand or reflect that demand for more custom solutions by
employees themselves and just for a more nuanced, reflective, personalized system.
Speaker 8 (12:37):
Yeah. So I would say, have you ever used open
source software? And have you used good open source software
and bad open source software? So I listen to my
podcasts on something called Antennapod, which an open source platform.
It's really nice. It's super customizable. It's also a little
bit of a pain sometimes because you can customize everything, okay,
(13:01):
because it's an open source project and people wanted it
to do all the things. And one of the reasons
why I like it is also that it's quite lightweight.
So this is a passion project. People have put a
lot of effort into making it do these things in
a way that is effective, right, because it's a much
smaller app than many others and it has all these customizations.
(13:21):
But that requires a lot of skill and a lot
of effort, right, So that's sort of my positive example,
And even that is quite confoluted, right, And there's a
lot of open source projects where you can see that.
It's just it's a complete manner, it's buggy, it's got
lots of overhead. It you know, doesn't have basic features
(13:41):
that you really want, and then it has so much
customization that it's completely impossible to learn how to use.
And that's why I think that it's really important to
have a concrete point of view and a united vision
even when you're trying to think about how to customize
thanks to.
Speaker 2 (13:57):
People, any other things you can see where you would
anticipate there could be an impact upon decks, the ability
to effectively manage and monitor decks in an organization any ways,
it could jeopardize the use, for instance, of next being
products or any variety of sm tools that you can
(14:18):
think of.
Speaker 8 (14:21):
Well, when I think of how it could have a
negative interaction with DEX, I think part of that is
just you know, the dangers of shadowy Tea. The security stuff.
A lot of the stuff Tim talked about immediately keep
to my mind in terms of it's very hard to
monitor or manage anticipate issues, and especially some of you
(14:44):
who did previously work at a cybersecurity company, one of
the things that is very true cybersecurity is that it
is much much harder to defend because you're constantly trying
to find every single hole and you can never find
all of the possible holes. Of looking holes into your
systems at random at an alarming pace, that makes it
(15:08):
really really impossible to do security.
Speaker 3 (15:10):
Well.
Speaker 8 (15:11):
I do think the other side of dex application to
sort of the IT perspective, right, So I think that's
also a totally separate question in a way, is okay, well,
how can vibe coding help people who want to automate
processes who want to have custom solutions and technical resources
(15:34):
that they don't currently have, right? And I do you
think that's both a cool application, right? Like I do
think it's a possible positive venue for democratizing coding in
a way, but I do think that it needs systems
in place to manage it to avoid all of these
issues that we're highlighting.
Speaker 3 (15:56):
I love it we're moving inside out.
Speaker 2 (15:58):
I think everybody be styled with all the cautionary elements,
and we're moving towards the positive visionary usages.
Speaker 3 (16:06):
Megan, bring us all the way there.
Speaker 2 (16:08):
If you would put putting some of these flaws to
one side, what ways do you envisage or are you
hearing about from the customers you speak to so often
that the vibe coding is beginning to help individuals or
deed organizations.
Speaker 9 (16:23):
Wouldn't that be funny. I can't say that I've talked
to anyone in our customer base that is using vibe
coding at the enterprise level, and I think that that's
probably due to the cautionary points that we've made so far.
I think scalability right like what Oreana just mentioned. And yeah,
not to totally just not non answer your question, but
(16:47):
another aspect of this that I was thinking about, obviously,
this vibe coding thing. I'm not a coder at all,
so I wouldn't I don't think I would ever try
it because I have no way of knowing if what
the code it spits out as useful or not. I
think about my own use of these Genai tools, and
I use them right to do some to draft copy
for me to do writing to. Basically, that's kind of
(17:09):
the only thing I use them for is to write
stuff for me. And what I've learned from working with
them for the last however long, is you have to
be really specific and what you want. If you just say, hey,
write me an article about this thing, it's going to
spit you out the most generic, boring like unreadable exactly
with the word count you ask for article with all
(17:30):
the creativity stripped out of it. And you have to
be very want article that covers this thing with this
audience and this tone in voice, cover X, Y and Z.
It's even better if you can give it an outline
and then it will you do something.
Speaker 2 (17:41):
But still, I.
Speaker 9 (17:43):
Mean, there's a whole thing like justice for the M
dash right, like nobody uses m dashes anymore because it's
such a sign that you're writing was written by Genai
and not by a human being. I think human beings
bring a creativity that Jeni just doesn't replicate. And I
think we've talked about this before, and I'm not a coder,
so I can't say whether that's happening with the vibe
coding as well. If it's just sort of the dumbification
(18:03):
of code output, like the way that it's become the
dumbification of writing output. The things that Jenai writes, they're
just not that inspiring, they're not that fun to read.
They're useful in certain contexts, right, like a marketing copy,
like a landing page or an email copy or whatever.
It follows a rope, standard set up, and so it's
okay that it's not like the most thrilling and exciting
(18:25):
thing you've ever read in your life.
Speaker 3 (18:27):
But I don't know.
Speaker 9 (18:28):
I think the beauty of having a human solve something
is that we come up with a creative way to
solve a problem. And it's that's the thing that I
think our customers get excited about with decks is that
it requires creative thinking to solve problems that previously it
didn't think they could solve. And now they hear a
pobly like, actually, you know, because we have this dex
data and we have this solution, we can solve for
that vibe coding, I don't know, maybe someone can will
(18:53):
hear me say this and come send me a message
on LinkedIn with all the different ways that vibe coding
has enabled them to be incredibly creative, and I would
love to hear that, but to me, it seems just
like another example of like we're just dumbing down the
output that we're getting out of these things. And every
example I've heard of people using vibe coding has been
to make those sort of small individual projects. A website
(19:16):
that can generate a goofy image, an app that can
generate something kind of silly, or I've been reading articles
about criminals using it to hack into things and ransom people,
which obviously isn't what these companies want. So at the
enterprise level, I think we're going to bump into all
the things we've already identified in this conversation. And I
(19:37):
have not heard about any of our customers using vibe
coding for their own internal software.
Speaker 2 (19:44):
It can sound fun, though what you use for you
What did you use tim for your experiment with it?
Speaker 7 (19:50):
I don't remember which tool it was.
Speaker 3 (19:53):
I'd have to go back.
Speaker 7 (19:53):
It was it was a few months back, but it
was it was very similar to Meghan, what you were
talking about, where the result that I got was very simplistic,
It was not as robust as I needed it to be,
and I really felt like pursuing that avenue to get
something better was going to be a waste of my
time because it just wasn't going to get there that
(20:16):
I would be able to do it much better. And
that goes to the skill of the person asking the question.
Maybe if I wasn't an ROI kind of a person,
I would have pursued it for a lot longer, I.
Speaker 3 (20:26):
Think, Tom.
Speaker 7 (20:27):
It also plays into some of the things we've talked about,
does a I really make you more productive or are
you chasing something that feels more productive but in the
end you're actually wasting more time.
Speaker 2 (20:37):
Interesting, and I should add, you know, congratulations to vibe
coding for being the ost Dictionary Word.
Speaker 3 (20:43):
Of for year twenty twenty five.
Speaker 10 (20:45):
But I will allow Sewan to finish this conversation off
with a prediction Sewan as to whether vibe cozing as
a trend is about to come crashing down in the
new year or we've only seen the beginning of its rise.
Speaker 6 (21:03):
I'm gonna say we're only seeing the beginning only because Meta, Google, IBM,
and Microsoft are willing to put it into training modules
and see what it can come back with, and they've
done it before in the past, where there's been big surprises,
So if they're willing to tinker with it, let's not
rate it off just yet.
Speaker 2 (21:22):
You don't want to go up against their collective might
too strongly. There's some strong there, well, I think I
think there's certainly something there, and I think there's certainly
some some guardrails suggested by a panel today which we
would all do all do well to keep in mind
as we as we continue to experiment heartily with AI
(21:44):
and in.
Speaker 3 (21:44):
So many different areas.
Speaker 2 (21:47):
But look, we've just come out of our I would
think of our series of shows celebrating marking our two
hundredth episode as a not just as the Deck Show,
but as Reality Bites as well. We all contributed to
ever to reaching this incredible numerical milestone, and we're just
entering that that period of the year where things start
(22:09):
to wind down and we're going to hear you know,
people who have been listening to the show for years
know that this panel we come together, we hash out
the year it was twenty twenty five and our highlights
and our predictions the next year, and we also Tim
and I will also be talking to returning show favorite
Jeff Right to hear how his years, is that right,
(22:33):
might not even have known that.
Speaker 7 (22:37):
By now we can assume is the perennial year.
Speaker 3 (22:39):
Then yeah, exactly right, exactly right.
Speaker 2 (22:42):
So we hope everybody listening joins us to that, it
joins us to those episodes, and thanks everybody for coming
on to discuss vibe coding and I'll see you very soon.
Speaker 1 (22:56):
To make sure that you never miss an episode, subscribe
to the show and podcasts Spotify or your favorite podcast player,
and if you're listening on Apple podcasts, make sure to
leave a rating of the show. Just tap the number
of stars you think the podcast deserves.
Speaker 3 (23:10):
If you'd like to.
Speaker 1 (23:11):
Learn more about how next Think can help me improve
your digital employee experience, head over to nextthink dot com.
Thank you so much for listening. Until next 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. Follow now to get the latest episodes of Dateline NBC completely free, or subscribe to Dateline Premium for ad-free listening and exclusive bonus content: DatelinePremium.com
Are You A Charlotte?
In 1997, actress Kristin Davis’ life was forever changed when she took on the role of Charlotte York in Sex and the City. As we watched Carrie, Samantha, Miranda and Charlotte navigate relationships in NYC, the show helped push once unacceptable conversation topics out of the shadows and altered the narrative around women and sex. We all saw ourselves in them as they searched for fulfillment in life, sex and friendships. Now, Kristin Davis wants to connect with you, the fans, and share untold stories and all the behind the scenes. Together, with Kristin and special guests, what will begin with Sex and the City will evolve into talks about themes that are still so relevant today. "Are you a Charlotte?" is much more than just rewatching this beloved show, it brings the past and the present together as we talk with heart, humor and of course some optimism.