Thinking on How LLMs (&GenAI) have Changed Our Thinking

This is the 29th episode of thinking unthinking podcast

Divya: Hi, I'm Divya.

Kahran: Hi, I'm Karin.

Divya: Thinking thinking.

Kahran: And this is thinking unthinking.

Divya: And welcome to the 29th episode of thinking unthinking. Today, we talked about using LLMs in our lives and how it has changed how we interact with the world. We took a detour into creativity and how we think about it and think things that we have made, whether using GPT or by ourselves. And this was an impromptu conversation, and we hope you enjoy. So, almost always, when we record our podcast, we pretend that we are starting in middle of a conversation. But most of the times, we are actually having conversations specifically for the podcast. But this time, for the first time, what we have decided is that we're going to record a podcast because we were talking about something interesting, and both of us are like, this should be a podcast. okay.

How has Kahran's worldview evolved using AI tools

So the question I asked Kahran was, it's been a few months of using LLMs, both of the visual kind and of the text kind, and what has changed for him? How has his thinking changed? So, of course, there is one layer of processing wherever. What can I use LLMs for? What can I use chat GPT for? But how has your worldview evolved as you have been using these AI tools?

Kahran: sometimes when I'm trying to think of a word like, there's this phrase which I'm really going to struggle to pronounce because it's from the French, but in abiance, I believe it is. And, like, I could not remember the word abiance. Right. but you can just ask chat GPT. I was like, hey, there's a word I'm thinking of that means it's like they're waiting. It's like, the gifts are in Abby ness. Something like that. Something is pending, and it just was like, oh, it appears you're thinking of the term in ambience. It originates from the old French. And, like, right? Like, it's like, you know, it's so annoying when you have, like, one of those, I don't know, what do they call it? Like, brain worms. There's some colloquialism for it or something like that you're trying to think of and you can't remember.

Divya: Yeah, yeah.

Kahran: That's maybe almost like a more tactical level of, like, oh, you know, here's this thing I could do, and now I can do it differently, where now I can do it better. But I think the question you were asking was more around, like, how do you think differently after you realize that these tools are available to you?

Divya: M. Because I don't know if you have heard there's a statement that is used sometimes in design the tools that we design, design us?

Kahran: Oh, no, I've never heard that.

Divya: Like, you know, the way we use language, it affects how we think.

Kahran: I.

Divya: The way we design our buildings affects how we feel and how we think and how we relate to different people, the way we wear our clothes or, you know, use our tools. Like, all tools that exist, they alter how we relate to the world in some way or the other. And, like, I have started noticing some changes in the way I think about things because I've been using AI for some time now, and I was just curious, like, because I know that you have also been extensively using it in both, like, personal and professional spaces. And how does that. I don't know, like, what has that done to the way you look at world and problems and your life?

Kahran: It's interesting. I think. I think what it is, is I've started to feel like there's someone I can talk to about things who will be knowledgeable. It's interesting. Like, I had a whole conversation with it, about, like, how my phone, because I'm holding it with my pinky, is kind of, like, hurting my wrist. And I was like, you know, like, if I move to a heavier phone, like, what else would I need to do or think about to just redesign that experience? Cause I have a habit now, right? I have a habit of lying down, holding my phone in a certain way, and that weight is going on my pinky. And so it was just like, someone who also knows about ergonomics, also knows about habit changing, also knows about, like, how difficult it is to do these things, right? Like, you can have a conversation with it about, like, how could I do that? Would this intervention work? I, like, had a different conversation about how I feel like it's hard for me to digest fibrous foods. and then, I don't know, I had a different conversation about this type of surgery that someone had told me about that they were doing called a whipple surgery. And what does a life look like after that? And just what does that surgery mean? It was someone I could ask questions to, whereas a lot of times when you're doing that kind of research, it's a one way research. You can go find information, but you can't dive deeper into parts of it or clarify your understanding.

I was interested about the word kids slang and where that came from

maybe another interesting example is there was a little while ago that I was interested about the word kids slang and where that came from. And then it was really interesting to realize that it came from the romanian language. Well, it came from, like, this thing called thieves can't. I don't know if you're familiar at all.

Divya: Okay. No.

Kahran: so thieves can't is this way of speaking that was between, like, members of, like, the underworld or unsavory characters so that then law enforcement would not be aware of what they were saying. So something might be like backslang, where the order of letters and words would be reversed or they would rhyme something. So stairs might be referred to as apples and pears or, metaphorical slang. So kids comes from that. And a lot of the words in thieves camp come from other things, like Romanian, which, interestingly, Romanian has these hints of, connections with Sanskrit, which was really just interesting, right? Like, I went down this whole conversation with chat GPD, which was so interesting to me. And that kind of, kind of free form following of ideas, I feel like was possible in libraries, and then it was possible in Wikipedia, but it's possible in kind of a different way now because, again, it's a two way concept. It's like what was possible in a library when you had a great librarian to talk to.

Divya: Hm.

Kahran: But was not really possible on Wikipedia. And now it's almost possible again.

Do you feel like the world feels more understandable now that knowledge is more accessible

Divya: Do you feel like the world feels more understandable?

Kahran: I feel like knowledge feels more accessible.

Divya: M what's the difference?

Kahran: I think the knowledge was always there, but it didn't feel accessible. For example, I have a passing interest in military equipment. So over the years, I've kind of researched various military aircrafts and armaments and whatnot. that knowledge is accessible, right. I could go up and research and find the information I was looking for. M but it's a very different thing to be able to understand how people making decisions or people operating in a different scene or a different environment might have acted. So, for example, when I was asking chat GPT about peeves Kent, I asked it, could you give me an example conversation between two experienced thieves discussing a more novice, apprentice who has committed something like whose robbery has gone wrong. Right. And now it was able to kind of give me an example conversation that might happen between these two people and one that would be highlight for me how experienced people or people who are conversant in this way of style of communicating might talk, at a higher level than someone who is less conversant.

Divya: M. How are you going to find that? Normally?

Kahran: Yeah. A similar thing would be to understand how does a military commander think about what types of aircrafts to use for what types of missions, or how do you think about, how to position when you're trying to make a show of force? Those are the kinds of things that you can't really research. The knowledge is available, but understanding how people apply the knowledge is a different thing. And I think that has that. When I say accessible, it's understanding, not just the core thing, not just that this thing exists and people did this with it, but what was the rationale that led to that? And I think when you understand more of what led to something or the rationale or the reasoning, that's what makes it accessible. You're m not memorizing, you know, why?

Divya: So, I started trying to learn philosophy back in 2015. I want to say, for whatever reason, I just felt like I needed to develop an interest. And I tried to get through Coursera. I tried to read readings of philosophers, especially coming from somebody who had a hardcore stem background. It was so difficult to get into it because, like, my brain just kept screaming at me, especially as I, tried to go through, like, conventional educational routes. What is the point? Why are we doing this? Why are we doing this? And that's the precisely wrong questions to ask when you are trying to learn philosophy honestly, like, when you're trying to learn anything, but, like, more so for something which is as abstract as philosophy. And, then I found this podcast called philosophize this. And that was such a game changer, because he contextualized everything that a philosopher was talking about, especially, how would this be implemented in today's world? What would it make sense if you were to look at the world today? There was this other channel called, PBS idea channel. He also used to do something similar, and it's only, like, over time, just exposing myself to more of these contemporary explainer people that I started understanding. Oh, this is what philosophy is about, and this is why it's there. And now I can go and read more complex stuff. Now, I wouldn't fall asleep if I'm reading Camus. Or so I'd be like, okay, I can still do this. But in the beginning, it was just so difficult to get my teeth into it. And the way you were just describing it, it made me think that, oh, you don't need to wait for an idea channel or a philosophize this. If you want to get into something, you can actually get into whatever topic you want to get into, because while GPT is not as high quality, it is able to contextualize the problem statement for you.

Kahran: Yeah, I think that is very true. I think it's also, you kind of hinted at this, but it kind of depends what kind of learner you are and what you find challenging. So for me, I find deciding what to learn very challenging. Right. Like, I almost want someone to help me curate the set of options available. But once I've decided what to learn, if I don't have agency in my learning, then I will fall asleep. Right. Learning kind of information like I was using in my examples for you was very difficult for me in lecture style formats because I didn't have control over it. But in these kinds of formats, it's really great because I feel like I have, I'm choosing it. M but the first problem still, exists for me because that's not so much being solved by something like this. That's why structured kind of courses don't have the same sort of, inherent recoil for me that I think they might for you, because they helped me curate that. They helped me decide. Oh. Out of all the things I could be learning in financial economics, here's ten topics in ten weeks.

Divya: M. Yeah. I find, a structured academy of infuriatingly boring.

Kahran: Yeah.

Do you feel like you're using genai for some things within education

Do you feel like you're using genai for some things within, within education?

Divya: I don't think that I've used it as much for the learning dimension. Like, I did ask it to structure a course on graph theory for me and like teach me graph theory day by day, but I went till like day three and then I dropped off. I'll pick it up again someday, hopefully, probably, who knows? It's very interesting. I don't think I'm using it as a knowledge source. I think for me, AI has enabled making a lot more than it has enabled learning and understanding.

Kahran: Interesting.

Divya: So, like, for me, like the space of, I, didn't know I could do this before, and now I can. That is what AI is enabling. And it's very interesting, like some subsidiaries of this, which is why I initially asked you the question that I've realized is, I think value of an individual idea is just not as much in my mind anymore.

Kahran: Okay.

Divya: Like, I'm, finding it way easier to throw out ideas. I'm finding it way easier to just like, keep trying for more options, even in my own mind. Right. Like that. I don't know if you've heard this, but, like, Apple used to have this philosophy where they would generate hundred ideas for a thing. Even if they end up using the first one, they would still generate hundred. and I think, like, I do something, I'm able to do something similar more now because my brain doesn't just give up after the first, like, five ideas or first ten ideas. It's very easy to, like, sort of keep myself there.

Kahran: Like, could you give me an example?

Divya: so there is a part of, like, you know, asking GPT to generate ideas. Of course, it can generate a lot of options. That's amazing. All of that is awesome. So, for example, the clearest thing was recently Charu and I were discussing some stuff, and we were, like, supposed to generate puzzle ideas and we didn't use GPT for it. We were just having a conversation between the two of us. But I know from our older discussions, we would have generated maybe ten options in that space, and we ended up generating like, 25 just within that particular space itself. Like, oh, we can do it like this and we can do it like that. And even minor variations which would earlier have been discarded. So in a weird way, at least in my mind, the creator and the curator roles have been separated and the creator is allowed to exist for longer, and then the curator is allowed to, like, you know, cut things out for longer. Which also means that, like, I find it way easier to say, oh, this is a good idea, but maybe not for this thing. Let's just get rid of it, even if it feels really amazing. Like, I think one of the most challenging things, at least for me as a creator, has always been, you know, kill your darlings kind of a problem statement.

Kahran: Yeah, yeah, yeah, for sure.

Divya: Right? Like, ah, this is really nice, but I don't want to be so attached to it. I, maybe I would come up with something better or even if I had have, like, you know, really amazing five things. I don't want to keep five things. I need to remove many of those. There was only space for two ideas. I need to remove three of those. And now I'm just, like, finding it easier and easier to remove those three ideas without it having an emotional cost on me. So I think, like, rather than knowledge creation is the aspect that has enabled more.

You use GPT to generate ideas for your upcoming game

Kahran: So something you said in there really, like, caught my attention. You kind of said it really in passing, but you use GPT to generate ideas. Talk to me more about that. Is it like, you'll have a problem where you'll be like, oh, hey, GPT, we're trying to figure out this. I don't know if it's the game you and Tara are doing. We're trying to figure out this scenario for our game. help me. What could be 20 things the person could be doing before they get here. Is that how you would frame the question?

Divya: so we had to come up with a setting, and we knew that we wanted a setting in India, so we asked GPT, okay, give ten examples of, like, you know, an Agatha Christie styled murder mystery is taking place in this location, and it's based in India and what could be those locations? And it gave, like, some ten options. And then, like, give ten more. And give ten more. Then, charu and I had a meeting, and we figured out, okay, what are the things that we actually like? And we realized, oh, we like desert surprise. We are both from Rajasthan. Who knew that we would be attracted to Rajasthan as a setting? Like, we really did not. But as GPT kept giving us answers, we were just like, M this idea really seems very nice. but, like, we asked it for, a bunch of different options and sort of giving slight variations. So, like, you know, nature oriented settings or more urban settings or more like, you know, old location settings. And then some of the ideas that it came up with was like, you know, a haveli in Rajasthan and a, ghost town somewhere, something in runoff gut, something somewhere else. Something somewhere else. Like, there were a lot of different kinds of options, right? Like something in chirapoonji. I don't even remember all of the, like, 50, 60 options that it generated, but, like, it would generate something. Charu and I would have a meeting, and then we would sort of whittle it down. We would figure out, okay, what are the things that we actually like? And then so Charu was doing this part, and then she would go back and she would tell GPT, okay, we like this, we like this, we like this. We like these because of these reasons. And then generate more options. And it would keep generating same thing it did for, like, you know, characters, same thing it did for, like, you know, different stories. And, ah, ultimately, we have to sit down and we have to construct a cohesive narrative. We have to give personality and color to these characters. But one of the challenges, at least, like, for us, when we are trying to build something, is there are so many details that are needed. It's almost impossible to generate them from the get go. Especially as you're trying to think about, is this going to be good for interaction? How is it going to aid in puzzle building? Is this setting the kind of world that we want to create? And, like, for us, we would not have been able to take something which was like this mammoth complexity and still feel like it's tackle able by just two people making it. M. But because we don't have to, like, you know, neither of us are writers, so we probably can't create settings as rich just by ourselves, but we are good creators, so we can figure out, like, we have good taste, so we can figure out this works, and this wouldn't work. And even here, what works in an interaction kind of setup that would make so much more sense if we had five examples and we can tell, okay, this one would work better, and this one would work worse. And how much depth does an idea have? And we can evaluate. It's very hard to generate idea and evaluate its ability to be multifunctional, but it's way easier to see ten ideas in front of you and rank order theme by multifunctionality.

Kahran: Interesting. Yeah. That reminds me of something that happened for me kind of early on in my experience with GPT. I have a friend who I'm trying to write a fantasy novel with, and we were really kind of stuck on a novel system of magic, and so we tried to get GPT to generate systems for us, systems of magic. And Kyle, my partner, is really just disheartened, I would say, because he was like, these are not novel. They're just like an amalgamation of other people's magic systems. And I think what occurred to me while you were talking there that I think at, the course, nothing is really novel, right? Everything is based on something else. I think what we do as humans is if you can break it down to the small enough elements, you can find just innovations on those small elements. And then when you put things together, you can put them together in a novel way, and it will feel completely different because you iterated those small elements, and then once they were built again, you built a different structure. And I think the thing that we ran into when we were trying to do, our GPT, driven creation is we didn't break it down to a small enough level, asking it to generate a full system of magic. There's so many components that, right, like, what is the consumables? What are the limitations? What are the outputs, right? But if we had instead, maybe broken it down and been like, oh, give me ten ideas for things that could be consumed by an interesting system of magic, right? And then maybe done 50 of those and figured out which ones we resonated with and constructed it from there, I think we would have had a much better outcome. It's interesting because a lot of the dialogue I see among creators, it doesn't really talk about this aspect. Right. Which is that you can take those components, that sometimes, like you said, there's so many details, and if some of them have gaps, it feels really weird to the reader, to the person engaging with your work. But if you can have GPT help you in evolving some of those components, especially the ones that you maybe don't feel as strong about, you can make your final output so much stronger. I feel like that's really been missing from the dialogue so far, at least, that I've seen.

Divya: You know, how when assembly line was first developed, it would be, or rather, when machines first started getting used in industrialization, one would think, oh, but the machine can't actually produce a full blown sewing machine. No, it can't. You really have to break the sewing machine down into the thousands of components it has and then create a machine for each one of those components, and then for the assembly of all different things that need to be assembled. I think that, like, as creators, we probably don't have that much deep insight on all of the things that are going on in our mind. Like, we have been working on this proposal for somebody. Right? And one of the things that I'm realizing is I'm having to articulate a lot of the things that go on in my mind automatically as I'm trying to explain to you, okay, this is what I'm thinking, and this is why I'm thinking. And some of it is also, oh, okay. I didn't know that this is how I would have placed these things because of these reasons, but in my mind, it would just make sense because it has worked. And it's almost like this weird heuristical system that now I have to almost decode.

Kahran: Yeah.

Divya: I also think that it might also be that, like, you know, you guys are really good at building a world, and so it would feel subpar. Like, I have found, at least in the creativity process, the LLMs are good at bringing you to the average level. So if you are really bad at something, you're going to feel like, oh, I got uplifted, and if you're really good at something, then you're going to feel like, this is a downgrade. I could do way better than this.

Kahran: Yeah, but that's why your point about breaking it down really stuck with me. Right. Because I think just to stay with my system, a magic example. what we got stuck with was we didn't want something that felt too powerful, and we wanted something where it felt like the consumable or the limitation was, real. Right. So both like, no one's, like, throwing fireballs, but also, like, you know, whatever you're doing, maybe it takes a long time or something if it has a big impact. Right. but then where we started to fall down was in kind of the, like, the onboarding, if you will, to use a tech term. But, like, how does someone progress from? Is it innate skill? Is there training involved? Like, what is the flow? Because I think what happened was we had kind of snapshots. We had a snapshot of certain characters, and we knew where those characters wanted them to be, but then we were trying to fill out the background of how did these characters end up with the skills and powers or whatever that they have.

I think driving narrative cohesion was the part that really felt hard

And then also, how do we make this into a cohesive system? M so I think if we had maybe broken down the parts of it that felt hard, right. And I think driving narrative cohesion was the part that really felt hard and then tried to maybe work on just those parts, we would have figured out that there was some part in there that we weren't as good at. Like you said, maybe we're really good at world construction, but there's some part in there that we're not as good at. Like, I was just saying, maybe it's the cohesion between the narratives. And then I think maybe what you do is you ask something like GPT be like, hey, we have these two characters. They exist in the same world. Can you help me with ten different backstories of why they could have the powers they do or something. Right. To be able to really drive into that part of it? That is the issue.

Google Translate is very good at translation. Have you used it for something like poetry

Divya: so one of my friends has been trying to learn French using GPT, and GPT is just, like, structuring different lessons for it.

Kahran: How cool.

Divya: Have you tried to use it for something like that?

Kahran: Not really. I've used it for some translation. It's very good at translation. It's really interesting because I actually, used it to translate some poetry from Turkish. And it was the kind of thing where it's so hard to translate online, especially poetry, because. Right. You're going to get, like, an exact translation from Google translate, which is not what you want, and that's assuming the language is in Google Translate, which isn't true for all languages. So that's been an interesting thing to kind of learn. but, no, I haven't tried to do, like, structured learning like that, where I've only asked it to translate in different ways. Like, song lyrics have also asked it to translate for me. M it's very cool. To ask it, to translate it, because you can, again, go into so much detail about it. Right. You can be like, oh, what is the context of these? Why would they bring up this kind of tree versus a different tree in that time period? And it'll be able to answer for you.

Divya: Oh, wow.

Kahran: Right? Like, it'll talk to you about the trees. In the 16th century in Turkey represented this.

Divya: Oh, wow.

Kahran: It's amazing, frankly, that has.

Divya: Oh, my God. Yeah. That has just inspired me to go and, like, look at some haiku through m, the lens of GPT, because I think, like, haikus tend to do a lot of that where there is a lot of contextual language that is hidden within those, like, few syllables.

Kahran: Yeah.

Divya: Like, because especially when you think about Japanese or Chinese, the characters can mean multiple things. But hopefully, like, GPT would actually know that this character represents this thing and this thing and this thing. And this is why this haiku is multiple meanings all in one.

Kahran: Yeah, that is interesting. Actually, I was looking while you were talking about haiku just now, and I found the poem. I asked. so it was actually a kurdish turkish author, and I'm not sure if it's in kurdish or turkish. but to give you a sense. So the first line, erugran valar Akhmis. This, is not what GPT's saying. likely refers to the blooming of the Judas trees, also known as Uruguan. In Turkish, their blooming symbolizes the arrival of spring. However, with the second line, the poet expresses mourning or sorrow, perhaps implying a dissonance between the beauty of spring and the emotional state of the speaker. The sense of sadness or mourning persists despite the blooming of the trees, perhaps indicating a personal or societal issue that overshadows those, the usual joy of spring. The next two lines translate, there is no spring and there is no country. These lines can be expressing the disillusionment or loss. The absence of spring might be suggesting a lack of renewal or hope. And the absence of country might be representing feelings of displacement or alienation. so it continues. There's about three more lines of analysis on the poem, but I think it's really interesting, right? Like, you wouldn't be able to kind of understand these levels. It's kind of like getting to talk to someone who actually understands turkish and actually understands poetry.

Divya: You know what? As you were just describing it, it just made me think about. Oftentimes when we think about solutions, we try to think 99% is as bad as 0%, but in reality, even 50% is so much better than 0%. Like, we would never be talking about a turkish poem, especially not in the depth of whatever this person is, like, you know, experiencing. And now just thinking about whether I can, like, you know, read something in Malayali or, actually, I don't know how good GPT is with indic languages, but, like, assuming that it's okay.

Kahran: Yeah. I mean, Gaurav used whisper, which is. Whisper is GPT's voice to text. So he used it for Bangla, which is what they speak in West Bengal. And, yeah, it was great. Got all of whatever he was trying to say.

Divya: Oh, wow.

Kahran: And that was diction, too, which I feel like might be harder than translation.

Divya: Yeah. And definitely diction is harder than translation. Okay, great. So I know that there is a lot of literature in India which is written in Malayalam and in Bengali, and it's inaccessible to me. I am not somebody who understands languages easily. They are one of the things that even if I want to try and learn, it's, like, really difficult. This could be, like, such a pretty way of interacting with a much broader spectrum of literature than I would have ever imagined before.

Kahran: Yeah. And I think there's something really special in being able to ask more. Right. Like, if I wanted to, I could ask more about those kinds of trees. right. Or ask more about metaphor, other poets who use metaphors of spring in turkish writing. Like, there's so many different paths for you to explore further, which in a lot of ways doesn't exist in most presentations of art. Also, there's some interpretation someone's given you and that's, you know, whether the curator's written a note or someone's presenting it to you. Like, that's. That's how much information there's available. And maybe you get a little bit more, but that's what it is.

GPT helps you find names that reflect your family's heritage

Divya: This is so true.

Kahran: Just because you reminded me when you brought up indic languages a moment ago. one other really interesting use case was my sister just had a baby, gosh, six days ago. And we were trying to figure out, a good name. And something I did is I told GPT all of our names. I told it my name, my sister's name, my grandparents names, my parents names, my brother in law's name, everyone's names. And I asked it to suggest names for us. M. And it did an interesting job. It came up with a set that was, like, reflected both families heritage. A set inspired by my sister's side, which is a Punjabi seeks a set, inspired by my brother in law's side, which are, south indian, they're the millions. so it's just interesting, I think that kind of thing is so hard to curate the massive amounts of information available and even to maybe see the connections, and it can just see connections in a slightly different way than we would. And then, of course, its ability to curate information is unparalleled. It's so hard for us, to curate information like that.

Divya: This is so interesting because, like, as you were speaking, it also made me think about a lot of the arguments against GPT or other LLMs are that, oh, it's just regurgitating information and it's not true creativity. But it's like, I find that question as pointless as, when people say, oh, but there is no true meaning in life. And it's like, sure, you can try and be cynical and philosophical, but you are living day to day. Sure, it might seem like, oh, but this is not true creativity. But if somebody helped you put into words exactly the thing that you were feeling, or helped you understand exactly the thing that you were trying to understand, then it doesn't really matter. Right here, for example, if your sister could find a name that reflected both of your family's heritage, then it doesn't matter if it is the name that only. Only your nephew has, or if it is a name that multiple other people also have, because, like, it is perfect in your context.

Kahran: Yeah, that makes sense. I think you might be conflating a couple of ideas there, though, because I think part of it, in some of these things is people feel like it should be a struggle. And if it's not enough of a struggle, then it feels like the place we've reached can't be the end place. You know, like, we may have been fine with getting, I don't know, a 60% if we had to work to get the 60%. But if the 60% came easily, then, you know, we're not fine with 60%, then we need at least an 80%.

Divya: I maybe. But I also think there is another thing. There is this aggrandization of specially creative skills since the Renesa period. where there is like, oh, there is a lot of struggle and there is a lot of this and there is a lot of that, and only then you get to be a true creator. whether it is like, you know, people going to museums, looking at modern art and being like, oh, my kid could have done that. Yes. The point of art is not for it to be inaccessible, that only a select few get to do it. The fact that your kid can do that is not a negative, it's a positive. Everybody should be allowed to do it. Everybody can make cultural contributions. Like, it's not a real thing because it didn't take effort. I think is a little bit of human psyche looking for a easy out. So that, quote, unquote, you don't have to put in the effort yourself. Like, if only certain people can paint gorgeously or write beautifully, then it's only their responsibility. It's not our responsibility to actualize those parts of ourselves. Then we can sit in front of the tv and do nothing.

Kahran: Like, I don't know, I feel like, again, you're conflating a couple of things, because I.

Many people think that most people are not truly creative

So I'm reminded, I think it was Nabal Rabbikanth I was listening to on a podcast, and he was talking about how he doesn't strive for happiness because he's realized a lot of these positive emotions are in, contrast to something else, right? You're happy because you used to be unhappy, and now because you move from that unhappy place, you're in a happy place. So he was saying, like, going for contentment was kind of the point he was after. But I don't know. I mean, as I think about that point in the context of art, I do feel people who have really felt strongly emotional places, their ability to kind of talk to that place and capture from that place, it's just more than if you haven't kind of had those kind of more extreme life experiences. So I do feel like there may be something about how more people are able to find resonance in something because the depth of emotion in something is so much more. Maybe this will be a crazy example, but I think maybe part of why, like, Taylor Swift has so much appeal in her songs is because she's capturing a level of emotion that so, like, people who lightly feel that emotion can resonate, also, people who really deeply feel that emotion can resonate. So there's a whole gambit of people who are able to kind of resonate. And I would argue that I think, like, kind of, maybe the great artists are able to kind of do that similarly. So I think that's one part. And then the other part is thinking about, I don't know. I mean, maybe as someone who struggles with, like, a fear of rejection and fear of being judged, like, I feel like a lot of people come from that place, and when you put out there stuff that's very true to you or stuff that you have to explain or might fear you have to explain right. Whenever you have too open of choices, whenever you could do anything, then there's some reason why you did what you did. And I think that fear of having to explain that reason of, why you did what you did and how people look at you and how they might think of you can really hold you back. No one wants to look like a fool.

Divya: Yeah, I was, in that panel, right? Like, we discussed it a couple of, podcasts ago, and one of the persons who asked a question, his, thesis was, once in a generation, you will find somebody who is truly creative. Like, you will come across a Michelangelo or a da Vinci or a Picasso. The sad part is, I know that many people think that way, that most people are not truly creative. And I think, at least for me, fundamentally, that is where the difference is. I think almost everyone is truly creative, because that's a little bit like saying everybody can be a cook, everybody can be a, writer, everybody can be an engineer. Like, you need the right tools and you need the path. Would you be the top zero 1% of your domain? I don't know if hard work is the only thing, or showing up is the only thing that is required, but I also think that, like, saying, oh, there are only this few truly creative people, and everybody else is just, you know, doing whatever is like, such a fundamental misunderstanding on a cultural level about how creativity works. Just like, you know, if you ask me to write a novel or I ask you to make a painting, you would be like, m okay, that's not where I feel comfortable, but that is a skill thing. That's not a creative thing, right? Whether we are creating relationships or conversations or jokes, even, those are creative acts saying that, oh, I made this and I'm gonna take it to the world and show it to other people. That part is scary, of course, but I think a part of it is coming from my irritation at the way people are thinking about, oh, but are LLMs truly creative? And are they not? And it's just like, is it enabling you to make more things? Because if it is, then who cares? Like, it's such an uninteresting question to ask. And I feel that way about humans also. Are they making things, then they are creative? And why do we box ourselves and others into these things that it's like, oh, no, but you're not. Only very few people get to be.

Kahran: That's, hm, interesting. I feel like. Yeah. That maybe the right model for thinking about alums is much more as an evolution of the existing tools we have. Right? I know that there are some writers who refuse to use computers, right? They use typewriters or people who like to write by hand. And if you think about it like those evolutions of creation all gave a certain framework of seeing things. Before there was automatic spell check, you could have more creativity in your spelling, right? But now, every time there is a, mistype or something, right, like, you're going to have an extra thought about it because of that system, right? So similarly, if an LLM model is going to propose a solution, you're going to have an extra thought about it because of the limitations of that system, right? So I don't know. It's, if you think about it as an evolution of a creation system, then, you know, no one thought a pencil was creative, and no one thought a, typewriter, ah, was creative. And even once, a, computer had autocomplete and autocorrect, no one thought it was creative, right? So I don't know. I mean, if you think about an LLM as being like, yeah, now, instead of it autocompleting one word, it's autocompleting a number of words. I don't know, it doesn't feel, it doesn't ring fully true to me, but I feel like it answers part of it. I'd have to ponder it a bit more, I think.

There's an interesting kind of gatekeeping around creation and making

M but as to your other point, I think there's an interesting kind of gatekeeping that happens around creation and making. and I think it's funny because I think it actually came from trying not to do that. Right. Like, maker spaces were designed to be open to everyone, and a lot of the vocabulary around the creator economy was the idea that, you know, anyone can be part of the creator economy, but instead it kind of became like, well, if you followed a traditional path, this is not for you. If you went to business school, you cannot be a creator. And if you, like, studied economics or. Right. Like, if you did any of the things that people do, you're not part of this. This is for everyone who wasn't part of those things. I think it's interesting now how people have started to look at themselves as creative, creative or not, whether or not they're part of the maker economy or not part of the maker economy. Because I would have thought of myself as a creative person, but not someone who really created until, I think, we started to work much closer together a few years ago. And now I don't really have those kind of concerns in my head anymore of like, am I an artist? Am, I a creator? Right. It just, like, feels much more immaterial. But for a long time, that felt very material. You know, it felt like, oh, you know, I'm trying to claim something that I'm not entitled to, which is a lot. What you were saying earlier, right. That we've created this myth of saying, you must have suffered in these ways in order to be the kind of artists that we should pay attention to culturally. Yeah, it's interesting. I don't know. I think that part of what was happening about a decade ago when design thinking was really coming to the forefront, there were a lot of books that I remember coming out. Like, there was a book called the Creative Habit. And these ideas around, like, creativity was a muscle you had to grow. I don't really know what happened to that, because I feel like when I was starting my career in the late two thousands, early 2010s, there was a big focus on saying, like, you know, yes, creativity is a muscle. You must focus on that and, you know, do small things, and someday you'll become more and more creative. I feel like that kind of dropped out of cultural parlance over the last five, three years. I don't know.

Divya: If you just think about it very objectively, it is more financially beneficial to somebody who is in the know. Like, let's say I have made my career as a creator to gatekeep. Yeah, it would be financially better for me if I gate kept.

Kahran: Yeah.

Divya: If I just told people, no, you are not allowed to be creative. Like, one of the debates I had with one of my friends a few months ago when we had not actually yet started using mid journey as extensively as we do now. And their argument was, oh, creativity has been open to everybody for so long anyways. And I was like, no, not everybody has seven years to dedicate to their artistic practice. I'm sorry, it requires a certain kind of privilege to say that, oh, you should just draw 3 hours a day for seven continuous years until you get better. No, not everybody has that kind of time, that kind of skill, that kind of money. Like, very honestly, we don't. And why should some people's ideas be worth less? Because they don't have the privilege. And again, access to these tools is also not equitable, but it is way more equitable than people who have only than just a small segment of people who have potentially spent some amount of time and gotten somewhere.

I think as cultures evolve, we look at ourselves in certain ways

Kahran: I think so much of it also has to do with what you think people like you do. You know, it's a little bit of a leap but I was listening to, I think a news story recently about how much, how important it is to give dignity to people. especially when there's been a natural disaster or something. And so you're having, like, you're giving help. I think it was about Maui and Mahaina and it's like. But you want to be doing it in a way that you're treating people with dignity. I was just thinking about that, that it's interesting how culturally in the US, we have so much of a focus of, like, you know, if you are going to, like, get, dishonored or something, it's better not to do it, even if it's like, kind of a question of like, shelter and, like, food and such. Right. Whereas I think culturally, in other places, that doesn't exist as much. I mean, sure it's there, but it's not as much that you need to show your dignity for all people. Right. It can be that, you know, there's a certain group that you're maintaining that for. And I'm thinking a little bit more of India at the time, but. Right, but, and, you know, as long as you, you're confident in your standing and youre group, it doesn't need to be for everyone. So, I don't know. I think as cultures evolve, we look at ourselves in certain ways and we deem certain things are accessible for us or not accessible for us. And I think in such a, like a dignity kind of focused place. Yeah. The idea that you would do something undignified, right? You do something, create something that a child could create, it's like shit. That's not. I don't want people to look at me like that, like in the us. And I wonder if it's true in other places. There's this whole wine and paint movement where you come. It's like a thing you do with your friends and you come and you drink wine and then you paint and it's like, well, you're drinking wine anyway. Like no one's going to expect your painting is going to be great. So it's taken this thing and given you almost an excuse.

Divya: It's so entertaining. Like the social structures that we set up and then the mental gymnastics we go through so that those social structures do not stop us from doing the thing that we want to do.

Kahran: It's very funny. And I wonder if we might get to a point where it's like, you know, created with GPT or something and that's like almost a badge on certain types of creations that lowers the mental barriers to doing it right. We know if it's curated with GPT or something, right? Like, people are going to be like, oh, GPT had a big part in that. It's fine. People are producing thousands of those a day. Who cares? You know? And it becomes this lower barrier of creation in some way.

Divya: M. Yeah, I'm excited about, like, how, I don't know, maybe less so now than I was before, because I can see that, like, people are already getting set in their old previous ways of thinking and, like, stopping to engage with, like, the difficult questions that these LLMs were posing. but hopefully eventually we may face those questions again of, like, what is consciousness? What is creativity? Where does value come from?

Kahran: You know, I think also what's happening is people are finding a use case and doing it right. Like, I have. I found a knowledge use case and that's what I've been using my LLM for, right, where. And that's why I think hearing you and now I've been able to imagine in my head where are different places I could use it for a generative use case, right. Where I could kind of, like I was saying, break down a problem small enough that I would love to know 50 different ideas around that or that part of the problem that didn't feel accessible to me as much before, right? Like, it felt like that was a thing that, like, other people knew how to do, but I didn't have those kinds of needs. And I just, like, I didn't think in a way that I would use GPT for. That was kind of how I thought about it. But now I'm, like, seeing there is a way. I don't know, I think when things are too broad, and again, this is particularly for me, but when things are too broad, it's overwhelming. So you're trying to make things into an understandable, digestible box.

Divya: Box.

Kahran: Yeah, yeah. And I think, like you're saying, right, those big questions of, like, what is consciousness? You know, what is creativity? Those are big, big, big box questions. no one wants to do that, but this is really cool. I'm glad we talked about. Yeah, yeah. Because I feel like now I have some things I might try out with GPT, which is good because I feel like, like I was saying, I think I've gotten into a particular style of engaging with it and I can even see my usage and started to drop off recently because I've been like, oh, I know what I can do with you. No longer fight 15 chats a day.

Divya: I'm also really glad that, like, we spontaneously decided, This should be a podcast conversation, and then we just decided to have one.

Kahran: Yeah.

Divya: I think at 29th episode, it's a good one to have.

Kahran: And I think this is, in my mind at least, this is one of our goals, right? Like, I think we've had so many interesting conversations over the years that that was what, in my mind, started why we should have a podcast, you know?

Divya: Very true.

Kahran: And so I think capturing one of them, it was cool.

Divya: Yeah.

Kahran: Okay, good job. Bye.

Divya: Thanks for listening to this episode of thinking on thinking. Our, theme music is by Steve Gooms.

Kahran: If you found this topic to be interesting or have other topics you wish we would explore on a future episode, please reach out to, us at, ah, our website joyous studio.