405: Nick Mehta, Former CEO of Gainsight, on Agentic Overload, Machine Consciousness, and Leading...

And so I think what's tricky is people right now want to form this opinion on what's happening and we don't really fully know. So for example, if a company does a layoff proactively because they think they can eventually replace with the AI but they don't have the AI stuff ready and they had overhired in the ZERP era. Is that an AI layoff or not? I don't know. But all I can say is those people don't have jobs, right? All I could say is to the person who doesn't have a job or can't find a job, they could care less whether it's called an AI layoff or not. They just want to get a job.

I think for sure, although I feel like we need the whole episode for me to respond to that introduction. There's for Okay, number one, I love that we're episode 404 because of course on the web 404 is not found. So, I believe the answers to whatever questions you're asking me are not found anywhere. So, we don't know the answer there. Number two, definitely did not single-handedly create the CS category. My god, there's so many people both within Gainsight and outside customers, competitors, they all made that happen. Number three, I need to find some better fashion role models to be compared to on your list here. Like some more manly fashion role models, you know? I think that's like a little bit of a mix and diversity there, but I get it. And then number four, I love the took a mediocre idea turned into a billion dollar. This is just this whole thing is amazing. I love it.

That's great. Actual quote might be true. Yeah. Also, by the way, I'm not a good economist or mathematician. I'm mediocre at all those as well. So, mediocre ideas and mediocre specialties as well. So, I'm not a philosopher. Mediocre closet philosopher. I'm mediocre. Yeah. Nothing original. But I think the thing that is true, it's a theme of your whole event is I think that what I love and I think you do great and a lot of people do is make things deeply personal about like what they can learn about themselves. And that's the one thing that nobody could be an expert on except you or me.

You know it's I was talking so I'm dealing with this and by the way I think like it's not like AI is the beginning of brain fry right like a friend of mine who I've told you a little bit about deals with just like brain fry around kids and like just the context switching there right tons of parents can relate to that right there's brain fry around supporting elderly parents or million different ways that we could have brain fry community people friends but I do think that there's an ability for it to kind of scale to infinity with AI where yes, like I don't know how many browser tabs I have open, but also not just that the tabs are open because they've always been open like that for a while, but more like Cloud Code has work for me to do in seven different sessions and I'm stressed out that I'm here on this podcast not responding to Cloud Code's needs. And there's some kind of thing over the in the future of like it is instead of us directing it, it's directing us. And there's a lot to think about there.

Oh, yeah. Tabs and each of them are shouting at you.

We've never had that before, right? That compelling feeling to be always on.

You said something poignant in the dinner that was part group therapy and part dinner. Which was no matter what I do, I feel like I'm behind.

Um that unpack that. What makes you feel that way?

Oh wow. What is it? What's your first reaction when you hear that? I won't be the witness. Just well, okay. Let's go through the letters again. Give me each one.

Brittle. Yeah. I think that I don't perceive the AI as brittle as much as we are brittle. I mean AI is of course a little bit brittle. Yeah. But it's just getting so better so much faster. To me that doesn't occupy my mind space. But yeah, we're brittle and our ability to use it is brittle. Okay. What was the next one?

Oh, yeah. I mean, just the alert, just the feeling like the FOMO, right? The extreme FOMO. But, you know, I guess we need a new version of FOMO for agents, right? That's it. It's not you're not missing out in some like, oh, there's a party happening somewhere and you're not invited. It's more like your agents need more from you, right? And so, that is real for sure.

Oh, yeah. Yeah. That's interesting because I find that I can parallel process so many ideas but that makes my mind so fragmented. So for example like I have a lot of thinking about what I want to do next exploring some building some just fun projects thinking about some ideas meeting VCs and so I'm constantly building stuff in Claude Code in my case obviously somebody could use Codex or whatever and but it's just like oh I'm building that thing and yeah I forgot about that thing and there's this other thing here and I'm constantly context switching all the time.

Well that I think is an interesting thing where you're kind of looking at this machine or whatever we call it and we don't really know how nobody knows at I mean you can understand pieces of how it works and there's lots of great work to try to what they call mechanistic interpretability to understand parts of the LLM but the truth is like as a system level nobody really fully understands all of it and so there's some element where you're just like you're praying to the AI gods and saying hey please build this for me and hopefully I did it right and like you just do that over and over again. So you could argue that creates more stress or maybe it's like less cuz maybe it's like okay well I tried and it just didn't give me the right answer. Let me try again. Now sometimes I do get frustrated. And what I find very interesting is I get less frustrated with coding but definitely with chat. Like I remember this conversation I had with ChatGPT on the highway where I was like sometimes I'll like talk to a just like nerdy topic. I was asking some question about infinities and set theory and things like that and there's this really just kind of basic stuff about the fact that there's like many levels of infinity Cantor had this whole proof around it's forever ago 150 years ago and ChatGPT tried to imply that two different things were actually different levels of infinity was and it was wrong and I was like don't you even understand Cantor's set theory and Cantor's diagonalization proof and like who am I yelling at? Like who am I upset at, right? Am I upset at the history of all humans that have written stuff that turned into the pre-training corpus? Am I upset at the OpenAI people that did the post-training? Who am I upset at when I get upset at AI? That's unclear.

Well, I think that I mean basic psychology anytime you're angry about something or upset about something then you're touching something internally, right? So that for sure always like everything is like that. Um, so you know there's some I don't know what anxiety that was tapping into about me, but etc. The thing that's different though about social is you can have this belief that it's a human that you're upset at. Now the truth is some of those humans were actually bots. We all know that. But there's some belief that you're like a human on the other side here. There's no illusion that there's a human on the other side. And so and what's interesting is like we speak of ChatGPT this goes into philosophy which we could talk endlessly about but you know if I say think of you as Dan I think of a person with an identity that is unique and distinct from other people at least as far as we know and when I think of ChatGPT am I like is that the server is that like one of the many servers is that the actual GPU is it well a neural network runs across lots of GPUs is it part of the neural network I'm upset at right is it if there's multiple instances of ChatGPT are they all one is it all AI and so I think locating the emotion is harder when it's AI like what is that emotion you have with AI I think people will run into this and probably are when they think about things like relationships companionship with AI etc right where I'm not here to judge everyone do their own thing but there's this question of like what are do you have a relationship with and you could argue and you could go back to like solipsism be like well there's nobody but me anyways right but to the extent you believe there are other people normally when you have a relationship with other human you think oh there's that other human and we have some relationship right and there's a kind of infinite like regress like mirror kind of thing that happens but with an AI like who what are you having a relationship with and that's what's very interesting so much philosophy to figure out.

Yeah. Saw that.

Um gosh, is that even the right question to ask? Like maybe we need a new definition of consciousness. And even if we had one, like who cares?

There's so many questions that came out of that. And if you know, you know, X is a you're using X, you know, formerly Twitter is amazing because your feed is just whatever you're into. And so in my feed, the AI and consciousness debate is like 90% of the feed. And there's like violent opinions on two sides about it, right? And there's what some people call like the functionalists which basically say computation is about say consciousness is about the function it achieves. So if an AI looks conscious it is. And then there's other folks who believe that no there's some deeper essence to consciousness. For example, it's tied to a biological substrate. It can't just be silicon etc etc right. And so that debate is unfortunately it's like kind of irreconcilable. They start with like just different worldviews like with two different people with different religions which are both equally valid but like in their own reference frame, right? I saw his Richard Dawkins. Richard Dawkins was great at like very pushing out very controversial stuff, right? So you also have to kind of he's brilliant, but there's a selection bias, right? Like just read anything he writes about religion and God. He literally that's right good memory. And so I think that I think some of it I was like, okay, here's Richard Dawkins being Richard Dawkins again, finding something controversial and it fits with his worldview around not religion not being a thing and all that stuff, right? But when I look at it, I think goes back to what you just said there'll probably eventually be an evolved taxonomy or definition of consciousness. You know if you look at consciousness there's many many different sub branches and stuff but like one of them was this very famous David Chalmers like kind of a scion in the world of consciousness hard problem of consciousness which it's not about like can we build somebody that build something that looks and acts like a human but instead can we build something that has the internal experience of a human. Some people call that qualia, that feeling, the taste, the like sense of what the color red is, the smell of vanilla, right? And so that idea of the hard problem of consciousness has been like kind of this center of like a lot of the conversation of consciousness and is extremely elusive to then make quantitative. Right? There's some good theories on this. something called integrated information theory which is basically complexity of a system is what drives the level of consciousness it has which gets into this whole other area called panpsychism where does everything has consciousness in it putting all that aside I do think that AI is going to bring all these topics to the surface and cause people from different walks of life to have to come together and come up with new points of view not just like replaying all the old points of view but the new ones I don't know what they are by the way I'm definitely not smart enough for that. But I know that like right now it's just like this log jam of two wildly opposing ideas that don't have any consensus.

That's exactly right. You know, David Deutsch is this physicist very famous in kind of theoretical physics and he wrote this book called The Beginning of Infinity and he said like broadly human beings purpose is to come up with explanations for whatever we have whatever we see and then learn there more explanations to be had and then find more and that desire to constantly find more knowing that we will all leave this earth or whatever without everything being resolved. We know that right? Nobody has an illusion that they will know everything before things are over. And so but this idea that like the beauty of debate which I think actually we it's easy to get lost in a lot of the things we could lose in this transition and those are real by the way to really be thoughtful about them. But one thing we could gain is maybe more mental and calendar space for true discussion about the bigger questions.

Feels like it's easier to reconcile that. I mean, where do you come out on that? The relationship between consciousness and the self and then does it make the argument about AI being or not being conscious any different? I'll just say when you frame around the self. I don't have a precise answer, but I'll tell you what that brings up for me, which is the when people ask me what's the book that influenced my life more than any others. And I know I'm supposed to say some business book or something, but I always say Gödel, Escher, Bach, which is an extremely nerdy book by Douglas Hofstadter who is a professor at University of Indiana. And his book, which became like kind of within the CS world and stuff, it's reasonably widely read. I mean, my understanding is the number of people that start the book versus finish, there's a big drop off, but it's 800 pages of like math and parables and stuff, but the core idea in it is self-referential loops. He calls them strange loops, something that refers to itself and observes itself. And so the idea basically is that like is consciousness a self-referential loop? And the book's called Gödel, Escher, Bach because Kurt Gödel was an Austrian logician and he came up one of the greatest logic proofs of all time. Gödel's incompleteness theorem and that was a self-referential loop to solve it. And then the same thing in Bach's music and Escher's art. And so the idea basically is self-referential loops might be core to our experience of reality and the consciousness we experience may be the self observing the self, right? And we have a little bit of that experience where we kind of are like, wait, I'm thinking about myself, but then what am I thinking about? Who's thinking about that thing? And you can get into many layers. Whether you go into therapy or people that have done psychedelics, you can experience this like loop or sometimes step out of it if you're in meditation or psychedelics, you can see that happening from an outside perspective, which is really interesting. And so I do think there's some reflection thing both with a person to themselves as well as between people that might be some core source of meaning that could tie to consciousness. This is pure speculation though. Like this is completely made up. Like I don't have any scientific data. There's no randomized control trial. I'm just this is just like maybe more of a hope or like a desire for that to be true.