Saranyan Vigraham
Director of Engineering at Meta working on compute and AI infrastructure; teaches AI and ethics to middle schoolers at Khan Lab School.
Auto-generated transcript — lightly formatted, may contain errors.
Saranyan
I'll just try doing something. Then it will not work out. I want, can, I can work to iterate on the tool really well.
And there's a friction in the process by itself because you want to create something in the world. And you're learning something. There's a feedback loop where your work is teaching you something about yourself. You're either growing, you're diminishing, whatever it is. But
Bruno Marnette
Mm-hmm.
Saranyan
the feedback exists. And wherever there's a feedback that exists, think friction automatically exists, I think, because the feedback loops are
The intent of a feedback loop is that it provides, it either tells you something is good or not and you keep it trading, right? So I think there's a natural fiction baked in. But any place where there is no feedback loop, because you can't even respond to the world, let's say you
Bruno Marnette
Mm-hmm.
Saranyan
create some art and you put it out there, you don't understand art, you don't know what you're doing there, you have a bunch of likes,
Bruno Marnette
Yeah yeah.
Saranyan
right? Then you cannot do anything about it, you can never act.
Philipp
I'm Philipp Zahn, my co host is Bruno Marnette, and this is a podcast about taste and the creative side of AI. Today's guest is Saranyan Vigraham, Director of Engineering at Meta, where he works on compute and AI infrastructure. Besides his main job, he teaches AI and ethics to middle schoolers, writes about AI and the humanities on his blog, and he is working on various side projects on the intersection of AI, culture, and ethical technology.
Bruno Marnette
So hello Saranyan. Thank you so much for
Saranyan
Hey.
Bruno Marnette
joining us today. we're very excited to talk to you because you are a real technologist. You you work at Meta, you're in charge of serious kind of AI infrastructure, and in a way you're one of the people who allow AI to work at the scale is working on today. but you're someone who also does a lot on the side, especially
around education I was curious to start by asking like you know what what motivated you to spend time on on education? What what gave you the appetite for the education?
Saranyan
I mean, so I have two boys. I have a 14 year old who is a rising freshman right now. And I have a younger one who is 11, who is a sixth grade, just started middle school. And when my older one was in middle school, I started teaching them. in middle school, I started teaching AI and ethics, mostly because of the conversations I've had with them outside.
Two things happen,
Bruno Marnette
Mm-hmm.
Saranyan
I think, with kids. It's really black or white. AI is good or bad. And they're influenced by the people around them. And unfortunately,
Bruno Marnette
Mm-hmm.
Saranyan
a lot of people they come in contact with have no idea what AI is.
Bruno Marnette
Mm.
Saranyan
is lot of fear and hype that fuels some of the beliefs. And kids absorb that really well. And so
Bruno Marnette
Mm.
Saranyan
when we have dinner table conversations, when we just sit and talk about a bunch of stuff, they have this
feeling of, it going to really be bad? Is it going to destroy the world? And I've had these conversations with my kids, where
Bruno Marnette
Mm.
Saranyan
I can give you an example where my younger one, he is a very creative kid. He wants to be a film director when he grows up. He reads screenplays, he writes screenplays, he's always writing stuff. He's way ahead in terms of his reading and writing for his age. He reads Stephen King. He's only 11, right? But he just
Bruno Marnette
Yeah. Mm-hmm.
Saranyan
has this inherent passion for telling good stories.
His hunger is a lot, so he's not satisfied with his age level material. He'll keep reaching and reaching and reaching to a point where he tried to convince me to let him watch movies that are completely at R and above. And I'm like, no, dude, you're not doing this thing. And I cannot always filter things for him. So we
Bruno Marnette
Mm-hmm.
Saranyan
came up with this fact that you can read screenplays. That's fine. I reasonable. I'll still make one pass on common sense media to see what themes there are.
But it's OK. I'll break the rule. You can go and read Screenplays. Because reading means that you imagine it.
Bruno Marnette
Mm-hmm.
Saranyan
You control what you imagine. And that's actually a healthier way of going and seeking broader stories, I
Bruno Marnette
Mm.
Saranyan
think. So once it happened that when he was in fifth grade, he went and told his teacher that, I want to be a movie director. And she responded in a very interesting way. She said, all AI coming in, movie industry is not going to be there.
Why do you want to be this?
Bruno Marnette
No.
Saranyan
And he came home crying.
Philipp
wow.
Saranyan
And I was like, wow,
Bruno Marnette
Yeah.
Saranyan
this is such a, I mean,
Bruno Marnette
Yeah.
Philipp
Soul crushing.
Saranyan
have a problem with that answer for many, many reasons, but even keeping all that thing aside, I felt that you just, it's a very misinformed way of engaging with a young person. And
Bruno Marnette
Mm-hmm.
Saranyan
it doesn't really help anybody, right?
So there are all these kinds of experiences.
Bruno Marnette
Mm.
Saranyan
And my older one in his time at middle school, yes, had this blanket policies about AI to deal with once his teachers called him that, hey, you you use ChatGPT to write this essay. And he was like, no, I did not write, use ChatGPT. he was pretty upset about that. So the problem is these are very black or white scenarios where this really does
Bruno Marnette
Mm-hmm.
Saranyan
not work.
I started engaging with his peer group to teach what AI is, what happens behind the scenes. And also I saw
Bruno Marnette
Yeah.
Saranyan
my kids hitting, my older one at least hitting that age where Instagram became addictive. Though my kids are not on any social media accounts, they're firm about that. That they're still not ready for any TikTok or Instagram or stuff.
Bruno Marnette
Mm-hmm.
Saranyan
But I had to explain him why that is the case. So I started building these
apps for them that taught them behind the scenes of how technology really works. What are these infinite scrolls do? Why do you end up staying on the platform? What does addiction mean?
Bruno Marnette
Mm.
Saranyan
What does engagement mean? Because at the end of the day, incentives are very different, right? Kids come for entertainment and peer validation, but platforms want you to stay there to make money. So there are very different incentives.
without even qualifying what is good or what is bad. They are persuaded into doing something without their knowledge of
Bruno Marnette
Yeah.
Saranyan
what they're agreeing to. And that's what I wanted to change. So I started teaching purely because I wanted to tell them, you can go, if you want to go on Instagram or TikTok, YouTube, that's perfectly fine, but know what you're getting into and every choice that you're making, there is a reason behind it. And as long as you know the reason behind it.
and you agree to it, then go do it, right? The moment
is
Bruno Marnette
Yeah.
Saranyan
when the beauty of this is they're generally very righteous about things. They care about the world. They're at the age, the middle school age is very interesting, I think. They do want the world to be a good place. There is that inherent passion for making things around you good. They get mad when things are not fair, right? Eventually they get desensitized.
So that was
Bruno Marnette
Mm.
Saranyan
the, that is the age where I thought, okay, if I go and teach now and we can have these conversations, so where we can preserve the sensitivity that they have with the world. At the same time, they also understand what is happening and then they have the agency to change it. So that's the only thing I feel you can give kids that we can give them agency, which means that we have to know, we had to also, we really cannot teach them anything, but we have to help them to understand how to think.
How do we look at something that they interact with and ask questions around what does it really mean? Right? Wow. And it's not somebody else's narratives you have to follow. You have to create your own narratives. And that's the only way to do it. Right? So that is the reason I got into this entire world just because I was selfishly wanted to make this better for my children. And then I realized that the broader effect is when their peers also believe in this.
So it's
Bruno Marnette
Mm.
Saranyan
better to not just do a one-on-one, but rather go in a setting where they and their friends are together. And you can have this conversation across the boards. And then I started building games for them. I started building a bunch of experiences for them. And it was really fun. So yeah.
Bruno Marnette
There's so much so much in there already. I love it. I I like that instead of trying to just convince your son to not go on TikTok, you decided to convince everyone around him. That's a pretty ambitious way to
Saranyan
Yeah, I think that's the way to do it, right?
Bruno Marnette
that's pretty ambitious, but actually more realistic because indeed it's all hard mentality, right? That's how we that's how we function.
Saranyan
Yes, exactly, exactly.
Bruno Marnette
and maybe one one thing I'm very curious about is when when your son came home disappointed about this.
this idea that, you know, the movie industry will disappear. What was your first kind of the first argument you brought in to kind of reassure him?
Saranyan
It was actually tough. had to think about it because first I had to just calm him down, say that, but he was super mad. So you know, it's funny. So when Google launched VO3, I said that, don't we just build something with it to see how the movie making works, right? And we did a bunch of prompts and stuff and they turned out to be terrible. And he was very happy with this. He's like, okay, AI can never replace movie making. This is the worst.
thing that like, you know, that I've seen in my life. So, you know, I'm not really worried about it. Right. So the fear went away
Bruno Marnette
Yeah. Yeah.
Philipp
Ha ha.
Saranyan
because he thought that, you know, this is just really cannot
do anything.
Bruno Marnette
That's really good.
Saranyan
Right. So that was, that was actually good though. That was not my intent. The outcome was okay. I think the intent was not that the intent was I wanted to wanted him to understand that tomorrow. If he, if he ever like, you know, when he, if he continues to pursue his dream and he has a list of film schools he wants to attend and he kind of, yes, his role models and all that stuff.
he can understand that, you know, I don't need $50 million to make a movie, right? I don't need to go and ask for money.
Bruno Marnette
Mm-hmm.
Saranyan
If I master the tools, I can probably do this with $50,000 my, like I can raise from friends and family, right? And that is a really strong place to be in. And I think that's what technology does, I think. It enables a of the, it enables and creates a lower barrier to entry for you to tell your story really well.
You still need to preserve the human element of it. And I think I have a lot. This is a place I think I have a fair bit of opinion on. But the tools will become so good, become so predictable that you can just use them to get ahead and tell the story of what you want. Because we always talk about this, right? Right on the shoulders of the giants. You always
Bruno Marnette
Mm.
Saranyan
like leverage. I look at technology as the same thing. It's a giant.
that you can stand on its shoulder and you can leverage, right? So that discrimination is important for children, I think. And that's what I've been trying to get him to do. It's still very hard. He still hates it. He still does not want to do it. But he's starting to warm up to the idea that maybe there is some kind of a...
there is some kind of value here. I can probably use something
Bruno Marnette
Yeah.
Saranyan
to tell a story, right? But I think it'll take him some time to get there. But at least the hatred is gone, I think. The fear is gone. And also, he basks in the incompetence of AI to tell good stories. And that is the viewpoint, Bruno, that you mentioned,
Bruno Marnette
Yeah.
Saranyan
where AI it doesn't have a good taste, right? So that is fine, I think. That works really well for us.
Bruno Marnette
I have a few questions, and so I have a niece also who's wants to be a movie director, and she's really angry at the AI industry.
Saranyan
Yeah.
Bruno Marnette
Part of it is copyright, right? There's something about the way
Saranyan
Yeah.
Bruno Marnette
AI came and stole everything in in the perception of many artists.
Saranyan
Yeah, of course.
Bruno Marnette
as you said, there's a counter argument of shoulders of giants. but there's well basically what I'm
What I'm curious about is how do you get people to kind of come to peace with how AI came about? Did you think there is something that that that could help?
Saranyan
I don't know man, this is one of the toughest questions I think, right?
I mean, we can see this. So it's not, it has happened in the past, right? Where we put things on the internet and people
learn from it, right? Humans learn
Bruno Marnette
Mm-hmm. Yeah.
Saranyan
from it. And we had no problem with it, right? And
Bruno Marnette
Mm-hmm.
Saranyan
we had the right attribution. would, there have been instances of plagiarism as well, but it was in a narrow case. You could always trace it back and you can, and the world was much simpler.
Bruno Marnette
Mm.
Saranyan
The
problem is the race to build the most frontier model in the world, people just swallow whatever is available on the internet. And now the question becomes, it's like this huge behemoth being that absorbs everything. It digests in ways that you cannot really figure out. It mashes it up.
there is a bit of an intelligence there that even though we don't want to call them emergent systems or whatever, but I think there is a bit of intelligence here because it is trained on the language of humans at the end of the day. And we always have expressed our intelligence in language, through language. So there is a bit of pattern and it can put things together really well. So the question becomes, traceability is very hard right now, right? What did the model train on is not open?
The second thing is, it using what part of its training set is it using to give you the result that it's giving? It's not determinable, right?
Philipp
Mm-hmm.
Saranyan
So the question is, we don't even know. We can always argue that this looks very similar to something that I've seen here, but how do you prove it, right? So it becomes more of a feeling-based thing than objectively
Bruno Marnette
True.
Saranyan
trying to show the weights of what it really means. That's one big problem.
The second thing is it suddenly changes the thing because now everybody pretends to be an artist. Everybody can open up a chat GPT, create an image, or use
Bruno Marnette
Mm.
Saranyan
Claude to build something. And even to a point where it's become so silly, even when I look at LinkedIn comments, you can see
Philipp
Ha ha.
Saranyan
that it's AI-generated. People just take something, put it on.
or chat our GPT and give the answer, which is just like generated, right? I'm like, what are we really doing? Are we just outsourcing the thinking? We want to appear smart. To what extent does it stop? Now, because of this, because internet is filled with people who want to create and people in the past, and and there were tears, right? And there were creators.
who were extremely talented. We followed them. We really admired their work and it was very inspiring. Now that entire layer is getting very polluted. everybody is a creator, even though they are not. So people believing that they're smarter than they are is actually a problem, I think. And also like the...
Bruno Marnette
Yes. this is a great this is a great segue to taste. Yes.
Saranyan
So
Bruno Marnette
Yeah.
Saranyan
then you lose that whatever genuity that you had there. Now it's just filled with like this massive volume of similar looking stuff
Bruno Marnette
Yeah, yeah.
Saranyan
and which makes it very hard for creators to really distinguish themselves. And it is frustrating and it's painful, right? And that kind of, that is the problem I feel that. So personally I feel it's horrible. I mean, it could be nicer if
Everybody at least attributes their work towards this part is all AI generated. I fall in this bucket of people who use AI to experiment and imitate and then we want to get better. And now this is my original work and this is my art. I would love to see that. You can use anything to get better in life, right? So that is perfectly fine. And then leave the creators alone. Let them just do their stuff. Let us
Bruno Marnette
Mm-hmm.
Saranyan
make sure that we...
attribute their work, we compensate them, we support them. And I think that's that tier does not exist right now. And how do you even create that tier where we are truly celebrating the creators around us, especially because in this AI generated madness that we live in, somebody creating something unique and individual need to be celebrated. They need to kind of, we just need to get behind them. We need to give them so much love, right?
Bruno Marnette
Yes,
Saranyan
But we're not doing that, right? And that's my problem. That's my problem. It's not from the content, but from what it is doing to this, like, the flatness it's creating in the world is my biggest challenge.
Bruno Marnette
The you reminded me a bit of the Yelp effect, right? There's there's a fantastic soft park episode about this on how you know Yelp came around and then immediately everyone was a critic. You know, everyone would go to restaurants and say,
Saranyan
Of course.
Bruno Marnette
Okay, I'm gonna judge the food. you're not gonna wait for the Michelin guide to come and you know evaluate you properly. You would just basically start having average opinion on on food and
Saranyan
Yes. Yes.
Bruno Marnette
I don't have a very strong opinion on this yet because I'm I'm not really a food critic and I don't know that world very well. But I would imagine a similar effect happen in many art forms where I mean obviously when Netflix come up with a new movie nowadays we know they optimize for a very different thing than the
Saranyan
Yeah.
Bruno Marnette
top critics, right? We know they now they're trying to make content that you can watch while doing something else because they know like half of the people are actually playing on their phone while watching movies, right? So there's
Saranyan
Yeah.
Bruno Marnette
all this kind of incentive for
for platform to essentially aim for the average listener, whereas we at least in some areas of life, they used to be maybe a more elitist, and this problem with that elitism as well, but we used to have
Saranyan
Mm-hmm.
Bruno Marnette
maybe said a more organized, more hierarchical methods to, you know, judge, you know, what what movie is supposed to be good, what food is supposed to be good and so on. Do you I guess it's more a common than a question, but I d do you think some of that is is is happening as well with
With with with technology in general? Like how do you see technology playing a role in that kind of first of all, i do you think it's the right way to think about this? It's like this the the end of this sort of hierarchy and elitism and then is there pros and cons to it? If I may make it a an open question.
Saranyan
Yes, I'll answer that but I want to hear what
what Philipp wants to say and then I'll answer this
Philipp
Okay.
I want to push back a bit. and I want to start with the example of a photograph or kind
Bruno Marnette
Mm-hmm.
Philipp
of the technology behind photographs. if you look at the 50s or 60s, basically, that thing was exploding. Now at that time we had a similar thing that many people were taking photos that were not able to take photos before just because of costs. So the the cost of doing that were dropping. I would still argue that in the 60s, sorry, after what 70s, 80s, 90s, and also up to this point.
the art form of the photograph itself still is there. It has changed, obviously. So clearly there is more of everything. But at the same time, I would also say there are this there are these peaks and there are more of them, and they're getting more in some sense diverse. What's your view on this?
Saranyan
Yes, I agree with it. I think both are very related, isn't it?
Philipp
Absolutely. Yes.
Saranyan
I don't think we will ever lose. So my point was not that we will lose that craft. My
Bruno Marnette
Mm-hmm.
Saranyan
point was that we have to somehow separate that craft and give people to give more visibility and support to the people who.
who love the craft and who are still honing the craft.
Bruno Marnette
Mm.
Saranyan
So this goes back to Bruno's point around, I have a lot of average, the food critics, right? The median has shifted, isn't it? In the past, if you draw a bell curve, the people that who were in the mid-band of it, that the median distribution
Bruno Marnette
Mm-hmm.
Saranyan
were people who took pride in that craft. They like, you know, created this.
whether it whatever artifact that they want, whether it is reviews, whether it's photographs, movies, right? But now the curve is shifting. I feel that the median distribution now is about people who have access to the technology, but they have not spent any time holding the craft. They spend some time with the chatGPT or any kind of AI tool that
makes them, that gives them the perception of being an expert, a pseudo expert. And then they churn out things. So that is where the distribution, the bell cover shifting
Bruno Marnette
Mm.
Saranyan
to, right? But then you have this long tail, which is really the people who pride in their craft. And do you know the story of Jiro Uno? Like, he's the sushi chef. Like, do you know this thing? Yeah, it's a great story. So Bruno, in a very high level is this person,
Jiro Ono, he has a very small sushi restaurant
Bruno Marnette
Mm-hmm.
Saranyan
near Tokyo subway and he's right now 100. He's an amazing, like, you he's
Bruno Marnette
Mm-hmm.
Saranyan
a very inspiring guy. He started making sushi from he was nine years old and he started, and that restaurant he eventually created, like, only seats 10 people. And it is a hole
Bruno Marnette
Mm-hmm.
Saranyan
in the wall, no frills.
Bruno Marnette
Mm-hmm.
Saranyan
It's a three Michelin star restaurant. you see about like, they, and the interns, they spend 20 years trying to make an omelet before they even get a chance to make a sushi. And that's intact. There is this, if you read his interviews and articles that have been made about him, he perfected his craft so much that he breathes his craft, right? And so
Bruno Marnette
Mm.
Saranyan
people like him will still stand out in the world. And we need more people like Jiro, right?
So now the question is, all I was making, Philipp, is that if that is now a long tail and it is not your median distribution, how do
Bruno Marnette
Mm.
Saranyan
you even get those guys to, how do we change the narrative to focus on those guys, right? And that's
Bruno Marnette
Mm.
Saranyan
my main point. And so don't have an answer to it, obviously, right? So the L previews, I think, is a good example. I think the problem, though, is
We also get very much conditioned to believe in those
Bruno Marnette
Mm-hmm.
Saranyan
reviews. We get how many times that I open Reddit to see what is the general consensus of something like that, right? Before I even try to spend
Bruno Marnette
Yeah. Yeah.
Saranyan
time watching a movie or something. And I've done this and my son gets really annoyed with me when I do this and he wants to read something or he wants to a movie
Bruno Marnette
Mm.
Saranyan
and I quickly look at.
Common sense media are ready to see like because I don't have the time to watch that movie and give them that feedback And he's like no you cannot just base things on common sense media or I think because there is a different way in which I look at things and this is not my representative audience right and
Bruno Marnette
That's right. Yeah.
Saranyan
and that is the problem and I do not know what and now we have so much information so many people putting stuff out there and putting content there is very easy and And then what how do you how do you even? filter
what is your representative viewpoints that you want to get from this entire collective intelligence which might not be the right sample for you.
Philipp
Can I so I think we're totally aligned and I I I totally get your point. It's really a different question. If there is much more of anything, even if the pockets are very good, the question is still how to find them, right? And how to also then attribute value to them. Now,
Saranyan
Yes, yes, yes.
Philipp
there's actually there's a bit of an irony on the example, because I think there is there was this documentary, Jiro Dreams of Sushi. This was the the starting point basically.
Saranyan
Yes, yes, yes.
Philipp
And if I'm rem if I remember correctly, the makers of this of this documentary actually went on to work for Netflix for the the Chef's Table series.
Which actually is taking this idea of we celebrate in some sense, you see where I'm getting in some sense. There's
Saranyan
It is. Yes. Yes.
Philipp
I would my point is a bit like I I fully get your issue of in in its music, it's it's extreme, right? You can be as good as you want, no one
Saranyan
Mm.
Philipp
listens to you. And then it's really the question of how to actually be recognized. and I think it's actually also a question of technology. It's obviously cultural appreciation, but also like what makes you know, what is a good way of pushing people into the front that serve to be that deserve to be in the front.
In some sense. And it's I mean
Saranyan
Mm-hmm. Mm-hmm.
Philipp
it's a very open question, but the specific case is just funny because I mean I I was recently in Sicily and there is this one cafe. It's it's horrible. There are so many people
Saranyan
Hmm.
Philipp
and they all go there. I mean it's nice and it's probably very good, but they all go there because it was featured in one of those one of those documentaries. Like totally crazy.
Saranyan
Ha, yeah. Yeah.
Bruno Marnette
Mm-hmm. Yeah.
Philipp
Bruno, please.
Bruno Marnette
Yes, so
I'm gonna make a segue here. It's in so we're talking about cr deep crafts, right? We're talking about people who find
Saranyan
Mm.
Bruno Marnette
time to develop I guess expertise in making things and that that includes also forming taste, right, in the in specific domain.
Saranyan
Mm-hmm.
Bruno Marnette
personally one of the concerns I have about the next generation end the user AI is that in the while they're making things, AI can sometimes be a shortcut, right? You can give them
Quickly
to a point where because they're not engaging the material maybe as long or as deeply as they would have otherwise, they may not have the occasion to actually form a very clear sense of taste. Right. one reason I'm mentioning this is we I don't know, we've talked for instance to people in law firms and they're very worried that the juniors there, every time there's any like, you know, t legal question, they just ask ChatGPT for advice and then make a decision.
Whereas you know,
Saranyan
Yeah.
Bruno Marnette
seniors they have been through you know years and years of this sort of like training
Saranyan
Hmm.
Bruno Marnette
where you know they really had to engage with the content and kind of form their taste. so it sounds that one one possible way we I guess we help younger generation to avoid this sort of trap could be I I guess first of all is I will ask you if you think the trap is real, like if there's a really a risk. And the second question is gonna be.
you hinted a solution already. You talked about this sort of intentionality, right? Like you want people to use technology in a way where they kind of understand what it does to them.
Saranyan
Thanks
Bruno Marnette
but but a pot a potential concern there is that people, including adults, very often do things that which they know theoretically are bad
Saranyan
Of course.
Bruno Marnette
for them, right? Like we all know we're wasting time on specific platforms and at at
Saranyan
Hmm.
Bruno Marnette
any age, right? knowing exactly
what it does to us and sometimes we do it anyway, right? So I yeah, I wonder on both fronts, like do you think there's a risk of I guess students being robbed of that opportunity to form their own taste in different domains? And
Saranyan
Yeah, yeah.
Bruno Marnette
and do you think warning them about them is sufficient?
Saranyan
Yeah, I mean, it's a super interesting thing because, see, I'll give you an example from my class itself. So we were researching on something
Bruno Marnette
Mm-hmm.
Saranyan
for advocates. And what AI can do, I mean, I'm going to use AI in a very broad way here, so bear
Philipp
Mm-hmm.
Bruno Marnette
Mm-hmm.
Saranyan
with me. So what AI can do is it can expose children to far more information very easily.
Bruno Marnette
Mm-hmm.
Saranyan
Even
Bruno Marnette
Yeah.
Saranyan
forget the hallucination part of it. We'll assume that these models get accurate over time and for factual things, they are not going to lie. This is a big assumption that we want to start with. In that case, it can expose them to more architecture, literature, music, design, arguments. It can even compare against eras. It can reveal what influences you have. It can challenge
have you challenge someone's argument, it can have a discussion which they might not find from another peer or another adult. So
Bruno Marnette
Mm-hmm. Yeah.
Saranyan
it's very smart, logical brain that can work with you based on your own range. And it can even increase your range if you honestly engage with it. So the problem, though, when it becomes a problem when
when it kind of shifts from being this design or brainstorming partner to something that starts choosing what's good for you. And this is where I've seen it to be a problem with children at least, is where it'll tell them, this is the right way to do something. Out of all this, my preference is this. This is where it gets, and then the problem is then it starts,
It starts to introduce an atrophy in terms of the exact judgment you're trying to develop. Because now it is not just presenting information to you and allowing you to come to and pursue your own path towards rationalizing what it is, but rather it's also trying to decide for you. Now the challenge here is the compressed time, isn't it? What does it take to build judgment? What does it take to... And I grew up in an era where...
I didn't even have access to technology. And I formed my opinion based on the people I've talked to, like going and playing out in the playground. I didn't have streaming. I got my first computer when I was in my 10th grade, and that was a very incapable machine, where I could write basic programming. And I would play a couple of 8-bit games. This was the scope of what I had.
So there was a lot of wonder for us because we saw what we could do and we tried to interact with the machine. We tried to write simple programs and stuff like that. So
Philipp
Yeah.
Saranyan
there was a lot of friction and there were lot of perspectives from people from different cultural contexts, different local contexts that they had. And all these things fed our journey into
forming an opinion about something, right? It takes time. Now it's not that, now everything is instant. I see this thing with my kids, right? They, if they want something and they expect that Amazon will send it to them in a day, it is, this is the world they live in, right?
Bruno Marnette
Yeah, no, that's absolutely true.
Philipp
Yeah. Yeah.
Saranyan
And anything beyond a day is not acceptable. Anything, sometimes I do this thing where I don't order from Amazon, I order from eBay and it takes a week to come and I intentionally do this and they get mad.
How can this happen? Why would you not order it from an Amazon? But this is just you need to break the conditioning a bit around what it takes. So then the question becomes, how do we really introduce time as a factor where you have all this information presented to you. Now you're not going to get anything for the next one week. You go and
use this time to build your own knowledge. And I think that's a very hard thing to do because everything is on demand these days. There is no concept of waiting for something. And so then how do you make it natural? How do you make the journey itself natural? How do you make the pursuit natural? So both my children, my older one went to this very small, small middle school Khan Lab School. It started by Sal Khan, who is the founder of Khan Academy.
and they focus on mastery-based learning. And it's a really good school because they don't have grades. They let you pursue mastery and you spend as much time as you want to get good at something. And my younger one now goes to the same school, just for the middle school, and my son moved out because he wanted to be in a bigger school for high school. But what the concept is, essentially that now you go and spend as much time as you want to be good at that.
is a concept that I think that's a concept I was going for in their education during at least their middle school years, which I believe are the most formative. now the question is, so AI is amazing. AI is amazing if you let it be a partner in learning about the world. And then if we can intercept, then they have experiences based on what they have and they can validate that experiences. For instance, if you're learning about culture, right?
and you learn everything about it, then go and experience that culture. Go immerse yourself in wherever you want. Go to Spain, spend a week there. Immerse yourself in the culture and see if what you have learned applies here and come back, right? And that loop is missing though. And how do we introduce that loop? How do we make that frictionless so that AI can truly be the superpower that they need to learn about the world because they have answers at their fingertips?
Now they use those answers to go validate their perspectives in the world and come back and even make the systems better by providing that local context back. So they make this personal model that works for them. That's why I'm a huge believer in personal AI and AI that runs locally on your systems. It's not owned by a frontier lab. It is something, because you don't need LLMs for a lot of these things. You need good reasoning models that can...
that can run on the edge, that can run on your devices, and that can run on your desktop secure. But it learns with you. And
Bruno Marnette
Mm-hmm.
Saranyan
it knows when to shut up. And then it nudges you to go and experience the world and come back. So in a way, I want that to be a parent's assistant or a teacher's assistant than a children's assistant. So I think that model would be amazing. And even if it will frustrate them, that's perfectly fine. But for that to work,
you need to make sure that they don't get the easy answers anywhere else, right? Because the world is full of easy answers right now. So I think that's take on it. And a long answer, right?
Bruno Marnette
You you touched yeah. No,
Philipp
Yeah.
Bruno Marnette
no, but you touched on I had a bit of a Eureka moment when listening to you. you made me realize scarcity is kind of important. So the you you talked about time,
Saranyan
It's important, yes.
Bruno Marnette
right? Time is very the Amazon effect is very real, right? I would say also the like people don't know how to think about what movie to watch, you know. I imagine your
Saranyan
Yeah.
Bruno Marnette
date night with your wife need to decide what movie to watch.
Saranyan
Right?
Bruno Marnette
People don't have the patience to like debate for 15 minutes what could watch, right? Anymore. They just
Saranyan
Yeah, of course, of course.
Bruno Marnette
next, right? and then and I remember from my childhood, music, for instance, was when I was young enough, I had pocket money, I had a finite amount of it, and I would go to the shop and I would actually decide what CD to buy, right?
Saranyan
Yep.
Bruno Marnette
and that scarcity kind of forced me to actually think about okay, I like this, I don't like that, and I'm I'm gonna buy this
Saranyan
Yes.
Bruno Marnette
CD, it's not this other one, right?
Whereas now you basically get a subscription to a service that gives you this abundance of listening
Saranyan
Yeah.
Bruno Marnette
to anything you want. And because you're not and incentivized to to to to
Saranyan
Yeah.
Bruno Marnette
make your own decision, your yeah, you you're not taking the time to form your own taste. I I I I worry, you know, pe people talk a lot about AI. Like one of the scenarios for AI maybe to bring us lot of abundance. maybe a drawback of that abundance is the same. It's
Philipp
I worry that we look for X1 I would like to pull our bundles.
Bruno Marnette
that actually we'll
Saranyan
I I agree.
Philipp
Okay.
Saranyan
It just makes things flatter, I think. I mean, that's my biggest problem with AI is not the, it's for all the awesome capabilities it can provide us. It's going to make everything, it does not, it does not reward persistence and delays, right? It's not, it doesn't, it has no.
meaning for those things. is an answering machine. You actually ask something, it will give you that answer. It really serves you faithfully in that thing. And as the models get better, you have the world at your fingertips. And the question is, why do I need to work hard for the same information? Is this a valid question? It's a valid question. You have a very capable model sitting with you. So logically, that's a valid question. it does not, tomorrow, think if Anthropic goes down or
OpenAI goes down for a week, the world will be in a chaos, right? People will not know what to do. And I always have lot of fun watching
Bruno Marnette
Yeah.
Saranyan
this thing when Anthropic has servered down. I go to Twitter and see people like rage rant about this one. I'm like, dude, but this didn't exist like one year back, right? But that's the thing. Yes, of course, of course, of course, of course.
Bruno Marnette
To make a to make a meta reference, people would call the cops, right? Which is what happened when Facebook is done. People literally call the police, right? Yeah.
Saranyan
So I think that's the real
Philipp
I didn't know that.
Saranyan
thing. It's a real thing, actually.
Bruno Marnette
It's a very I think
inside Facebook when there's a bad accident they call it called the cops level incident.
Philipp
I see.
Can I maybe okay, may I sorry, I'm an economist by training. Maybe let me let me try to give a bit of a different perspective here because I mean so in in the let me let me realign, you know, let me be the nasty economist.
Bruno Marnette
Please. Yeah, we agreeing too much by the way guys. So there's too much alignment here. So please. Yes.
Saranyan
Please, please.
Hahaha
Philipp
Maybe let's take back I mean, you know, if you depending on where you lived fifty years ago and even to this day, it might well be that to get drinking water you have to walk quite a while, right? This is a form of friction that, you know.
market systems, etc. The economic growth that we have experienced has basically overcome to a large degree. Not in all places, but in many places. And in some sense, Amazon is the extension of this. It's not different. It's just basically more of reducing friction that is individually valuable. now on the other side, I mean this is this has been predicted I think in the 70s by Herbert Simon. It's that, you know, in the end we will have so much things that the scarce thing actually is the attention that we need to kind of manage.
Right. I mean, we all know this. This is not new, it's also social media made this point very
Saranyan
Right? That's right. That's right.
Philipp
strongly. But you can make the analogy. So the question is basically if you if you reverse it, the question is where should we have friction? Right? Where should we explicitly slow down or not be able to rush? And for children, it's bas typically means that parents or schools have to set some boundaries and have to introduce the these frictions. And we know this as well because we go to gyms. we don't have robots basically waiting the I this is one of the analogies a guest made in
Saranyan
Yeah.
Philipp
our on our program.
We don't send our robot there, we go ourselves because we know the friction that we experience there is good. So maybe if we start with this, and if you take this as a kind of as a as a as a ball that I throw to you, and you try to see where would you, you know, how would you take this up? Where is the friction individually? Let's say mass make let's make it more concrete. Mastering a specific thing like mathematics,
Saranyan
Hmm.
Philipp
or pick your pick, you know, do you think and how we should have this friction? And how should
Maybe how could it even be supported by technology? You mentioned one where
Saranyan
Right.
Philipp
we are not connecting to the global models, but we have our own local models.
Saranyan
Right, so I think this is probably the best question to ask Philipp. Where does friction even make sense? You don't want to go back to the age where progress was, we didn't have access to the basic stuff and things were not safe and it was much harder. So I think definitely agree with you that technology has.
It's been amazing to advance humanity to a place where the world is more connected. People live longer. We have cure for diseases. All these things are amazing. Obviously, you don't want friction there. But I feel friction, the way I think about it is I always play the negative side of it. If at all.
Like, let's say that I'm doing something today with technology. Is it a tool, or is it an enabler, or is it my twin which gets something done? And meaning that if it's a tool, without that, I'll still find a different tool. might actually not be able to do it as effectively as possible, but I can still probably do it. If it is an enabler,
it is that I still like, you know, I might not be able to do it by default, but because it enables me, gives me to a different
Bruno Marnette
Mm-hmm.
Saranyan
playing field. I'm able to now accomplish
Philipp
Mm-hmm.
Saranyan
that one. I still have some agency. It's still my problem solving. like I can create something, make something with my hand. The last part is where it is very, very tricky. I think where I have no freaking idea how to do this, but it feels good. I've seen other people do it there and I admire them.
Now with this, can completely hack my way and I can pretend I can do that. I have an absolute problem with that. And so one and two, I love it. And I want, that's where I want friction. If like, if I can, if it can enable me to understand, ask these questions, it can like make me like, for instance, a better tool means that I have a bunch more challenges. I'll just try doing something. Then it will not work out. I want, can, I can work to iterate on the tool really well.
And there's a friction in the process by itself because you want to create something in the world. And you're learning something. There's a feedback loop where your work is teaching you something about yourself. You're either growing, you're diminishing, whatever it is. But
Bruno Marnette
Mm-hmm.
Saranyan
the feedback exists. And wherever there's a feedback that exists, think friction automatically exists, I think, because the feedback loops are
The intent of a feedback loop is that it provides, it either tells you something is good or not and you keep it trading, right? So I think there's a natural fiction baked in. But any place where there is no feedback loop, because you can't even respond to the world, let's say you
Bruno Marnette
Mm-hmm.
Saranyan
create some art and you put it out there, you don't understand art, you don't know what you're doing there, you have a bunch of likes,
Bruno Marnette
Yeah yeah.
Saranyan
right? Then you cannot do anything about it, you can never act. So I can give you an example. So I was working on two ideas, okay?
And one was in cybersecurity, which I was fascinated by. I only know it through what I have read and understood. Other one was in developer tooling, mostly around how do I get the models, AI models that I'm working on to align, not at a foundation alignment that the labs work on, but
Philipp
Mm-hmm.
Saranyan
I want them to align to my policies, my views. And I wanted them to, so if I were in an enterprise and I have my own alignment layer and runtime, can I ensure that the models
I can steer the models to work on
Bruno Marnette
Mm.
Saranyan
my alignment, right? And that one, I understand it really in and out, that space I know very well. And I have opinions on it. And I started testing both these ideas out. The first one, it turned out that I could create stuff very well because I understand technology well and I can come up with something reasonable and very convincing as well. But when I went and showed it to people, they had a bunch of feedback.
But I couldn't move past that. I needed other people to tell me what I was doing wrong or right, what works and what did
Bruno Marnette
I didn't say.
Saranyan
not. I had zero opinion on that.
Philipp
Mm-hmm.
Saranyan
And so, but the second one, I didn't care. I had a lot of opinions on this. I would show people, but there's some opinions would agree with me, some would not. But all of these things served me to go and come back and ask myself the question of what am I even trying to do here? Does this work? Does it
Philipp
Mm-hmm.
Saranyan
not work?
there was inherent friction and I continued working on it, working on it, working on it. And a bunch of like about 50 or hundred friends are testing that thing out and that product, really, really like it just purely because of like, you know, this approach, right? Yeah.
Bruno Marnette
The engagement. Yeah. Yeah.
Saranyan
So I think, so the answer, for me is I want friction. If I have an opinion on something
Philipp
Mm-hmm.
Saranyan
that has to change in the world, I automatically have introduced friction because the world doesn't exist.
to fulfill my fancies.
Philipp
Mm-hmm.
Saranyan
And now if I build something around it, then I can improve
Philipp
Your your yeah.
Bruno Marnette
Yeah.
Saranyan
the tools. actually, through that, I can also grow and stuff. On places where I do not have either the skills or opinion, if I engage and create something in the world, there is just no friction there. I cannot do anything there. That's not useful for me. So
this
Philipp
Yeah and
Saranyan
is how
Bruno Marnette
Yeah.
Saranyan
I would bucket it.
Bruno Marnette
Can I I I relate a lot to what you just said. I wonder maybe to which extent I relate because we're both engineers by training a bit.
Saranyan
Yeah? Yeah.
Bruno Marnette
basically I think my part of my engineering, I guess, mindset is that I tend to gravitate towards hard problems. Like if a problem is too easy, I just get a bit bored. And
Saranyan
Sure. Sure.
Bruno Marnette
if there's no, I guess, social feedback loop, if there's no users.
things end up just being like an evening project and then it they die. You
Saranyan
Exactly.
Bruno Marnette
know, like I I really need this sort of a sense of
Saranyan
Yes.
Bruno Marnette
purpose, I guess, right? and purpose cannot just be a number on the screen. Like it needs to feel it needs
Saranyan
Yeah, I agree.
Bruno Marnette
to feel a bit more user facing.
Saranyan
Yes.
Bruno Marnette
To be nice, I know people who don't even need the user feedback. You know, some data
Philipp
I know people have gone with a label because the fact that
Saranyan
Yeah.
Bruno Marnette
scientists they just love the data so much. They could, you know, they could spend months just digging into data set and they they
Saranyan
Yeah? Right, right.
Bruno Marnette
like the numbers, right? So so I think this variation into what drives people in that sense.
Saranyan
I agree, I agree,
I agree.
Bruno Marnette
but I I guess the commonality is purpose. Like there need to be some sort of
Saranyan
Yeah.
Bruno Marnette
motivators that make you want to do the hard things. Is that is that your mental model?
Saranyan
That's right, because even when I build games for my kids in the class, the joy really comes from, I don't know if I got it right until they play that. And
Philipp
Uh-huh.
Saranyan
so for instance, we built this game to learn about AI biases. What I did was I built a game called Who is the Hacker? Essentially a bunch of pixel characters from Dice Bear. one group of kids, there are two groups in the class. First group would train.
these random like avatars that came on the screen as could this be a hacker or not and blah, blah, And then the other group will try to reverse engineer like, you who could be the hacker, right? So what it taught them was it taught, for instance, a good example, good outcome there was at the end of this thing, one kid said that, my God, like I marked everybody with a hoodie and glasses as a hacker. And, but my friends wear hoodies
Bruno Marnette
Yeah.
Philipp
Yeah.
Saranyan
and glasses too.
And they did not get
Bruno Marnette
Very cool.
Saranyan
this when they were playing. There is a subconscious bias. It comes from you watching movies about hackers or reading comics and whatnot. But this is important. Otherwise, what I built is useless, I think. And because I can never, when you're building things for
the world, the world has to give you feedback. When you're building it for yourself, you have to use it and provide feedback honestly, right? The second part is much harder because we tend to be tolerant of our own things and we can convince ourselves that what we built is reasonable. But
Bruno Marnette
Mm.
Saranyan
I've only recently gotten to an honest discrimination where I can call BS on myself and I build something and I use it for a week or two. I don't convince myself that what I built is good.
I can say that this absolute trash, I just stopped doing this one and let's just bring something. So I think somehow that validation has to come back.
Philipp
I if you did
Bruno Marnette
I think there's
a sorry Philipp, but I just know that well, what you what you're touching on is I think culturally very important. I think we're gonna need a large number of people to develop this skill now because
Saranyan
Yes.
Bruno Marnette
there's just so many people right now who are wasting their time with Lovable, with all this you know vibe coding tool, they're making apps no one needs that they don't need. And they they haven't developed this sort of critical view of like yeah, reflecting on. Yeah.
Saranyan
Yeah, people
love to create. At the end of the we're all creators. We want to put things out in the world. And I think it's amazing. That's the best part about being humans, because we are born to create in whatever way. what AI, like what are all the lovable cursor, what it has done is it has given people an independent way of creating something. It's OK not to talk to people anymore. You can sit in. It's the same thing when you go to parties these days.
People are comfortable looking on their phone and like, you know, sitting in a corner, right? But
Bruno Marnette
It's horrible. Yeah. Yeah.
Saranyan
in the past, it was not like that. You would have to... So when I was in my PhD, so my professor taught me this thing. So he would take me to every conference. He was one of my biggest... He was one of the guys who had the most influence on me in my life. And he would take me to these parties. He would tell me that, you know, don't go to the talks, stay outside, talk to people during break sessions, during coffee and go out drinking with them.
That's the only way you make connections. Nothing happens in the talks, right? Don't go sit in the talks. And what he would do is, I learned a lot of tricks from him where he would peel the label of a beer bottle and he would cover his badge with that label. So it happened there was a beer called Skullsplitter. I don't even know. They are so vaguely remember this one.
Philipp
So yeah, you can't remember that if you drank it.
Saranyan
Exactly, right? And he put that on top of his badge and he kept, he walked around, right? And people, there was a conversation. Suddenly, we were like, okay, what is this thing? Right? And he was so good at making these conversations and he was all, so that is where the real things happen, where you talk to people. So one of the most amazing things that happened to me in the last year is that I got a dog. Okay. It was my kids convinced me to get a dog, a puppy. And obviously it became like, you my responsibility, right?
And he's been amazing for me because of him, I know everybody in the community, most of the people in the community that I didn't even know before. I've met so many interesting people. I know their stories. I hang out with people. I talk to them. When I go to the park, I just have a lot of these really good conversations about, and I know what people and their lives, right? And it is amazing. The moment you have that kind of a touch point with people, it is fantastic.
On the contrary, that is what building was all about. When we go to hackathons, we build things like 10 years back or five years back, it was a community thing. You come together as a team, sit here, talk about your ideas. You don't even know the person you're building the app with. And it was a social thing. You knew about them, you built something, you deployed, you demoed it, you won a prize or not, whatever, nobody cared. But it was an amazing community thing, right?
The problem now is with lovable, with cursor, from a community thing, it has become personal. It has become so personal that you don't even have to talk to anybody. You can talk to an AI agent and you can put it out there. So you're losing the texture of what you get out, which is a process of building something. The output is probably better than what you could have created, but it lacks that other side of it, other invisible texture that it paints.
the process of creating something. And that is the part that I don't want to lose.
Philipp
Can I sorry again? I'm I'm trying to push back a bit. so
Saranyan
That's okay, you should, of course.
Philipp
I maybe let's make a distinction because I mean I agree if you're building something with the intent that it has some impact. It has only
Saranyan
Mm-hmm.
Philipp
impact if there are people who are using it, who are enjoyed enjoying it, whatever. But
Saranyan
Yeah? Yeah.
Philipp
let's make a distinction. If you're a writer, basically, you are this is a very lonely thing, and it's basically your own
Saranyan
course. It is.
Philipp
evaluation that that basically is is key for
Saranyan
Yes.
Philipp
development.
Saranyan
Yes.
Philipp
Clearly,
very often these people also learn from others and et cetera, et cetera. And then I think there's the other side. There's building and there's playing. And what I see, I see adults actually playing in a way they have never played before since they were a child with lovable. And
Saranyan
Okay? Okay.
Philipp
I think this is something I I also get Bruno's point basically. There's a lot of stuff that basically is out there and will not be used. But my children are also using Lego. They're building a lot of stuff that has no intention beyond it. It's just play for them. And I'm
Saranyan
That's right. That's right.
Philipp
I wonder how to balance this basically, because
Saranyan
Yeah.
Philipp
There are
these two sides. In some sense, it's enabling and it's playful for at some level, but it also destroys something or it kind of endangers something on the other side, as you discussed.
Saranyan
I mean, as long as it's clearly a play and we should do it, right? So I used to write for middle grade children. So when my kids were growing up, I wrote a of stories. at one point, I had about four manuscripts and submission. But.
But I've learned the hard way that publishing is very hard because it's what the market
Philipp
Yeah.
Saranyan
wants is sometimes what determines what gets published, right? But it is a lonely process, you're right. And there is a lot of joy in creating something and putting it out there for yourself because it is a very gratifying thing. So I absolutely think you should have fun. You should actually have the smile on your face and like, know, and put it out there. Even if you have one person using it, your family, your son, your kids playing it. So many dads I know who build things for their kids to play and they're so happy, right?
And that is a great thing. I think the challenge becomes where you're trying to push it out and make money out of it. And everybody believes that this is the next big thing, right? And I think that's probably what Bruno is talking about as well, where it crosses
Philipp
Mm-hmm. Yeah, I agree. Yeah.
Saranyan
the threshold of being a personal joy-giving thing.
to something that has to live in the world and like thousand people have to use it and this is the next big thing. I think that's where it gets very cloudy, I think.
Philipp
I also think, I mean, there is this infamous or famous point of by Kevin Kelly and 1,000 fans, basically, that you don't need many and very often your own, your type itself actually is maybe a good predictor of the social
Saranyan
Yes. Yes.
Philipp
group that you need to find that actually would value this. But I totally also get, I mean, it's exact it it is maybe maybe this is the cultural cultural issue that we are still associating the technical hard part of building a software software app, which was still hard a couple of years ago, as something which has status.
Saranyan
Yes, yes.
Yeah.
Philipp
And
with the status basically is typically associated, well, these people will, you know, it will be needed, it will be bored, it will be used, et cetera. But that's obviously not true anymore if everyone can do it.
Saranyan
Yeah, think so. Yes,
yes. And also I think, you know, the changes the way software gets built, right? For tomorrow, I can really think of, I've been thinking about this for my kids, that what does a new operating system look like? Right? I can have a shim on his Mac, which is completely personalized for him. Apps get created on the fly. All I need to ensure is there is a secure layer and he's well protected and it is safe. But then nothing.
stops you from that Lego world where you create what you want for yourself and it keeps reshaping it as you evolve, as you change. So totally agree with you. And I think that's where we are heading, I feel, that things will become very personal from, and you don't have to cater to the common denominator of something built for.
Majority that you have to agree you can live in your own local context you say that this is what is for me And I'm going to I have a system that really understands What I want in my life and it actually optimizes it and makes it so efficient for me to get that It's an amazing thing. I think so. It's it is fantastic if we can get that I feel right it will it's another opportunity to like look at people's
whatever, for them to project their needs out in a much more tangible way. I think it will be fantastic if we can get there, but I think it will take some time, but definitely exciting.
Philipp
Ask you on what I found very interesting on your arc. I you started on a very technical level, also did the PhD. Then you
Saranyan
Yeah. Yeah.
Philipp
know you worked in kind of hard engineering problems, and you are still working on this. At the same
Saranyan
That's Yeah.
Philipp
time, you have this kind of I mean, you you just said you wrote basically four books. and then there's also this side where I mean I looked at this discovery your intellectual voice and these tools about basically in some sense that are almost about finding yourself, which is quite interesting, right?
Saranyan
That's right.
Philipp
Because
There are you know, culturally there have been only so many tools that you could use for this, right?
Saranyan
Yeah, yeah.
Philipp
now as a society we have we had the church or equivalent things, institutions.
Saranyan
Right? Right.
Philipp
What made you start working on these tools in the first place? And what's your goal with this?
Saranyan
I have no idea for them. So this is actually the
Philipp
Ha ha.
Saranyan
toughest question you can ask me because I feel I'm still figuring out what I want to do. I have no idea, right? But
Philipp
Mm-hmm. Welcome to the club.
Saranyan
I feel the only thing I can do is to be honest about what I don't want. So I believe in finding who I am by the process of elimination, because that is the easiest thing for me to do. I do something, I throw away, and I know that this is not what I want, and then I'll go do something else.
And so my wife keeps, mean, for not me that the amount of opportunity costs that I have lost by doing these things, tinkering and throwing, tinkering and throwing is insane. It has been, it's been insane, right? Because it does impact the people around you and she gets the brunt of it at the end of the day because of all the long hours I spend on something which doesn't materialize to anything else, but anything else tangibly. I would not say anything else, anything else tangibly, right? So, but, but my intent is to...
be the truest person I can be in this moment. And I love a quote by Gandhi that I'm not consistent. I cannot get him quoted, but what he says is, I'm only consistent to my thoughts today, who I want to be today. I'm not consistent to my past. I could have said something that day, and today I'm not that person anymore. I believe in it quite a bit because we change as people.
This moment, I want to be the most consistent version of who I am, what I'm saying, what I'm thinking. And if that changes, I want to be the most honest and consistent person of that person tomorrow, right? So I think that's the thing. It also gives me an audit rate in my life. I write a lot. So I know my own arc. And the reason I write is, see, my parents are not writers. My dad is 80 now. And my mom passed away about 10 years back.
And now I cannot go back and I go and ask my dad and mom about their life and like how it is. I really want to know my dad's life and he was my age and how, what he went through. He cannot tell me much, much, and he's not a big talker as well. So he will like, why are we talking about all this stuff? He's like, yes. Right. So the thing is, but I do not, I know for a fact that the relationship I have with my boys, at some point they will come back when they are 40, 45, they'll be like, okay, what kind of person was my dad? Right.
And at that point, I might not even be able to give them the answer. I don't know. I might be not as articulate as I am today. But whatever I've left behind becomes that audit log for them. They can go and read things. my dad thought about it this way, but he was absolutely wrong. And he kind of followed his thing. So that arc, and the only way to capture that arc is through all the artifacts I leave in the world.
And that's the only purpose of me doing anything right now. So tomorrow when my kids want to know about me, they know about me through the work I left behind, whether it is writing, whether it is code I create, whether the websites I have built, and whatever it is. And so my hope is in that process of doing it, maybe before I die, I will know who I was and what I was trying to be. Today, I don't know that.
Philipp
It's so it
it is so funny because my father is a was was an engineer. And now that I see you both talking and Bruno basically
Saranyan
Okay?
Philipp
nodding, I can suddenly sense there is something there that I that all then that now clicks about how my father what was what my father was doing essentially because he was exactly about these artifacts and going somewhere.
Saranyan
Yeah.
Philipp
by the way, it it just reminded me when you started out saying that I just finished a book, which is a small book by I just looked it up. there is a there was a sentence in there, it's which is kind of quite promised by kind of quite
Famous by now. It's it's called I don't know where I'm going, but I know how to get there, basically. and it's
Saranyan
Yeah. Yeah.
Philipp
it it is by what's his name? It's by
forgetting about. Boyd Boyd Varty It's a it's a book by him. It's it's
Saranyan
Okay, I will read it.
Philipp
it's it's a it's a very short book. It's actually quite interesting book because it's anyways, I don't want to say more about it. But it struck me because this is this is something where again where culturally we're very often conditioned on what's the goal? What's the fastest pass there? And again we're talking about the friction, right? Sometimes the the fastest pass is not the one that is actually good for you or good for
Saranyan
That's right.
Philipp
the people around you.
No question here. Sorry, it was just an observation. We like I can just res it resonates very
Saranyan
I don't know, makes sense.
Philipp
much on on several levels for what you said.
Bruno Marnette
Likewise, I was gonna make an observation rather than a question. I relate a lot with what you what you said in terms of reinventing yourself every day, discovering who you are every day. I have I have one of those personalities as well, where if people look at my CV, they will always think, my god, if you stayed in this company 10 years ago, you would be so rich right now.
Saranyan
Yeah.
Bruno Marnette
Like I used to work at Palantir before it became so controversial, back when it was very a very different company. and yeah, I lasted a year. And and this and and part of that yeah, I've I've had like this natural rhythm of almost, you know, every year, every two years be feeling like I'm a different person. and I I think there was really good reason in the past for for society to push against that sort of lifestyle, if you will.
Because I guess at the society scale, you need some people to be very persistent, whereas you need others to maybe be more the the creative type and try different things. but one question I have about I guess the the the world we're entering with AI is let's say a hope I have is that maybe this will become more normalized and maybe because opportunity costs of changing domains, being kind of multidisciplinary, I as I like to be, in a way
Hopefully it will cost less, right? because you'll be able to learn faster when you switch domain, you'll be able to
Saranyan
Yeah.
Bruno Marnette
adapt faster thanks to the the new tool.
Saranyan
Yeah.
Bruno Marnette
so I'm hoping part of me is hoping that maybe society would become kind of more tolerant to that or like more approve more of
Saranyan
Yeah. Yeah.
Bruno Marnette
that kind of lifestyle. Is is this the direction of travel, you think, or?
Saranyan
I think so. makes a lot of sense. think extending the thing that you said, it kind of makes us as humans
more adaptable, right? If we
Bruno Marnette
Mm-hmm. Correct.
Saranyan
develop a healthy relation with technology, it is a great superpower for us, right? You can actually, you can pretty much take any of us and put it in somewhere. And if you have enough curiosity, you can do amazing stuff, I think. And also it's probably the newer version of the wandering hunter gatherers, like where you just keep moving around.
Bruno Marnette
Mm.
Saranyan
And because that's how you're trying to find...
You're not doing it for survival anymore. You're probably a higher self-actualization goal. Probably that's what we're doing. So I think it's amazing. I think we should all do 1,000 things before we die. That's a more interesting life than sticking with one and playing it safe. think suits your point, suits for some people, but I like
Bruno Marnette
I that's yeah, yeah. Couldn't agree more.
Saranyan
that experience. really appreciate
Bruno Marnette
Yeah.
Saranyan
that.
Philipp
I j it just crossed something crossed my mind because we were talking about that you producing the artifacts and I was just thinking, I had this I have this habit of telling stories to my children, which they quite enjoy. Mo most likely because not because of the of the stories, but more because it's by me. And they say they realize it's kind of it's personal, it's Indra and individual.
Saranyan
Right? Yeah, of course. Yes, yes.
Philipp
I was thinking at some point about recording them, just for the sake of it. And now I was just thinking, I mean let's
Let's turn this around. Is there let's say everyone does this? We are we are kind of almost drowning in these artifacts of personal things.
Saranyan
Yeah.
Philipp
My first order reaction would be that's awesome. I mean, if you know, if historians in the futures are looking at us, they will understand us so much better, and by extension, also themselves. I mean, we have a hard time of understanding how Roman life was for the average person because we have no no artifacts
Bruno Marnette
Mm.
Philipp
basically, except from the the rich and the famous and the powerful. But now while I was thinking about this, are there any downsides for this? Are there any
From your point of p from your point of view, are there any arguments for maybe this is why this might not be a good idea for storing and having all these artifacts floating around?
Saranyan
I so I probably, I'm going to say something that is very strongly opinionated and not borderlining or controversial, I
Philipp
Mm-hmm. Perfect.
Saranyan
think.
Bruno Marnette
Perfect.
Saranyan
So I don't think having more information or less information makes somebody capable of understanding us better or less. That's not the necessary condition. I you can understand about
about the other person with one conversation. You can actually look, if you look deep inside yourself and look at your own reactions to how you are, to your world and the people around you, you can understand really pretty much a lot of things about people, about yourself. So it's not that lack of information is making us, like making us understand the world less. It is that lack of observation, lack of spending
Bruno Marnette
Mm.
Saranyan
time, lack of going deeper, right?
So I mean, yeah, if you generate all the information, you can create it. It's a great repository for your children tomorrow to go take a look. But who knows whether they will take a look or not? Maybe they will run a notebook LM on top of it, and they will say that, summarize
Philipp
Yes.
Bruno Marnette
Yeah.
Saranyan
everything my dad said. You never know. That's what's going to happen, because that's exactly right.
Philipp
one page that's too much. One paragraph is
Bruno Marnette
Yeah.
Saranyan
So yeah, so I mean, we can create all these artifacts, all the information, but people who consume it?
is essentially determine how that's used. Somebody can spend days and days going through everything you have written, and they spend like five years to understand who you are and try to get into your head. Or they spend five minutes. You do not have control over it, I think. So I think the real understanding happens when we truly, it's, we can just, you have less samples, but you spend more, you engage more with the samples that you have.
That is my belief. would love for anything that my parents have told me that very easy. I can write them down in like five pages. That's all I know about them. But when I spend more and more with those things that they have told, and it gives me more perspectives as I get older, as I see it, this is exactly probably what he meant. I go and talk to my
Philipp
Yeah.
Saranyan
dad about it.
So I think you
Philipp
Yeah.
Saranyan
can always revisit the same thing that
Philipp
Yeah, it's
Saranyan
you have with different perspectives. And so in that manner, the thing that you have, the limited artifacts you have are already infinitely expandable, right? And why do we need things that are know, infinitely more in scale?
Philipp
Absolutely. Yes. Yeah. It's you
can make the same argument, right? There's some if you reread a book, you will discover more. And there are some books which you can basically reread a lot of times and they are still giving you value.
Saranyan
Yes, exactly, Yeah, absolutely.
Philipp
Yeah.
Saranyan
Absolutely. There is this, on the books, I'd probably give one unasked recommendation. There's this book called Siddhartha by Hermann Hesse. It's a 1919, you know this? That's my favorite
Philipp
Yeah, I know it. Yeah. Yeah.
Saranyan
book. And I've read it like a million times already. And every time I read it, it gives me something very new about myself.
and stuff like that, right? So as you go through the journey, there are some books and some materials that you have aggregated in your life that give you company and they also change with time.
Philipp
Have you have you read other works by Hermann Husserl? Like no I I don't really know the German titles because I read them in German, Narziss and Goldmund.
Saranyan
No, I only
read Siddhartha but I have like wanted to read, there's one more thing that I wanted to read. The Pilgrim. Yeah, Sheff and Wilkes
Philipp
I that's the sh the Steppenwolf is the it's the famous one from the seventies.
Saranyan
is the one. I've not read that but yeah, but Siddhartha is something I'd keep a good book. I would love that. That's fine. Okay. Okay. Okay, all right. I will.
Philipp
If I can give you a recommendation, read I mean I have to look up the English title. It's Narziss and Goldmund in German. It's a it's a it's a very, very interesting book. Specifically if you liked Siddhartha then it could be a very good good book.
We've taken up a lot of your time. This was a blast. Thanks a lot
Saranyan
Thank you so much.
Philipp
for being part of this. This was really, really nice.
Saranyan
Of course, I loved it. Thank you. Thanks Bruno. Yeah, yeah, awesome
Bruno Marnette
Thank you so much, yes. That's excellent. Thanks.