Is AI Making Us Stop Thinking? Two Doctors Debate

Adil: Fantasy just drives me crazy. It drives me crazy because like I Adil: don't really care about it that much, but then I care about it Adil: depending on who I'm playing. So I'm just like I'm always behind. Yeah. Humayun: Yeah, some people care too much about it. Some people Humayun: care. Some people care a lot about it and you know, Adil: I know. Humayun: on the week to week I just I have contempt for Humayun: some people that I just wanna beat. And just for that Yeah.

Adil: Yeah, my Adil: my need to shit talk will make me do enough Adil: research just to have a starting lineup that isn't injured. Yeah. Humayun: Yeah, that's about it. Otherwise I personally Humayun: don't care. And you know, Adil: It's it's too much I mean it's too much to keep track Adil: of. There's just too much going on. I'm like, dude, I'm not gonna Humayun: Dude, there's too much in life to keep Humayun: track of. Just like what happened to me tonight. Adil: dude, you got owned tonight.

That sucks. Share share with the share Humayun: Yeah, I mean it was horrible. I mean Adil: with the fans your your grievances. Humayun: Yeah, the five fans that are listening to us right now. Huh? Adil: Dude, I don't think there's that many. Adil: I don't think there's that many. Humayun: Not even I Humayun: mean, I feel like we should just do an episode or Humayun: we just talk about nothing. And let's see who even comments. Adil: Yeah.

Humayun: Like even if you just like you know just just say random stuff Humayun: through this episode and not really mean anything or do anything. Let's see Adil: Yeah. Humayun: who all comments. Watch this episode go the most viral on Adil: That'd be funny.

know, that's that there's like actually a real scientific study Adil: that was done that way, 'cause these authors were tracking to see if the Adil: journal editors and reviewers actually even read the papers or it was more Adil: random in which articles got published. And so they wrote like a real seeming Adil: intro and a real conclusion, but in the body of the actual research Adil: papers they were complete garbage, like actual total farce.

And they even wrote in Adil: there, This is a test, this is a joke, like this is not real. Adil: Papers and they submitted it to multiple journals and they had like Adil: an over fifty percent success rate at getting it accepted. Meaning it Adil: was never read by even the reviewers or the editors of the journal. Humayun: I mean now you can you know, you can just have Claude or Humayun: ChatGPT write the whole thing and not even double check it and just Humayun: put it somewhere Adil: Right.

Humayun: and someone will accept it and publish it for you. I mean, Humayun: you know, I think the whole world is just now AI talking to Humayun: AI.

you know, on on LinkedIn you you use the AI automated Humayun: system to post something on AI and within thirty seconds you get a Humayun: reply from an AI bot representing someone else you know posting on Humayun: your on their behalf and then you know your own bot replies and Humayun: says something to them and it's just a bunch of AI bots Humayun: talking to each other on LinkedIn.

And that's what I feel like like Adil: You know, I I wonder Adil: how much of like you know, the internet in general and Adil: social media is just some version of a bot or AI just Adil: communicating with another version of a bot or AI. Like what percentage Humayun: Yeah. Adil: like I would be very curious to know what percentage of the Adil: total interaction and communication that occurs on the web of any kind Adil: is actually human to human versus human and bot or bot to bot.

Humayun: Yeah, I mean I think it's a lot and it's growing Humayun: exponentially and it it's it's going up and up because I Humayun: think people are losing their ability to to think and create. Humayun: and and even think on their own and you know on the on Humayun: the I mean, you know, a lot of the emails now that that Humayun: I get from from people, I can tell it was not written by Humayun: that person.

the issue that I'm noticing is I give someone an assignment, Humayun: you know, where we're cre have we have a project that we're implementing Humayun: and I tell someone, Hey, why don't you look this up and kinda Humayun: create a plan, you know, create a plan for this this project and Humayun: to how to implement it.

And next thing you know, they get it I Humayun: get an email, hey I've created this you know plan to Humayun: implement this and I can tell clearly it's a sixteen page Humayun: multi paragraph like long words, tables in there, Adil: Yeah. Humayun: you know, something created by Claude clearly. I Humayun: was like, okay, if if I really needed you Adil: Yeah. Humayun: to create that, I could have just gone on and Humayun: created myself. That's not what I was asking for. I was Adil: Right.

Humayun: asking for a short, simple plan for you to kind of Humayun: come up with implementation plan rather than send me, hey, just Humayun: read this Claude summary, you know, that you created. But I Humayun: think that's what the world is getting towards, you know, and Humayun: And I think what we're gonna do is we're gonna leave our f Humayun: forget our problems solving skills, you know.

Life is going to throw a Humayun: bunch of curveballs at you, and th the best way to learn how Humayun: to solve those things is to kind of delve in and deal with Humayun: problems yourself, right? If something comes your way, first you panic a bit. Humayun: Then you kinda s kinda collect your thoughts a bit. You come Humayun: up with a solution. You implement a plan to solve that. And Humayun: then you you you you implement the plan and you execute the Humayun: plan.

And then sometimes the plan doesn't go the way you want Humayun: it. So you kinda pivot. You know, just what happened to me Humayun: to this mo this tonight. I mean, you know, I was I was Adil: Yeah.

Humayun: I I just got a car last week, Humayun: a new car that I've been kinda, you know, Humayun: driving around and my son has been super excited about the car Humayun: and he's been telling people, Hey, we got a new car, he's Humayun: going around telling everyone and then, you know, I I've shown the Humayun: car to a few people.

and then we just went for a Humayun: family outing and then I'm we're driving back on the freeway and Humayun: all of a sudden I just see the tire pressure going, you know, Adil: Yeah. Humayun: thirty-six, thirty-four, twenty thirty-two, twenty-eight, Adil: man. Humayun: fourteen. I'm like, Lord, I'm on the middle of I-10 Humayun: And my car starts getting a little shaky as well, so I just Adil: Yeah. Humayun: fall over. And then, you know, it's like eight eight, seven forty-five Humayun: PM.

So my wife is like, it's bedtime. I'm like, okay, it's Humayun: bedtime. I'm sorry. Like the car just we just had a flash. No, Adil: Obviously obviously this is your fault now. Humayun: I mean like, yeah, it's it's not my fault. I mean, Humayun: you know, we just and then I panicked a little bit. I'm Humayun: like, okay, like the kids need to get home. I can't Humayun: just put them in an Uber because they need the car seats.

Humayun: You know, so you you have that sense of panic for Humayun: a second, you collect your thoughts and then you come up with Adil: Yeah. Humayun: a plan, right? So I was like, Okay, well, let's call Uber, Humayun: let's transfer the car seats to the Uber. Let me call AAA Humayun: to get my car fixed while you guys go home, you know. Humayun: And then and then the the the AAA guy shows up and Humayun: he's like, that spare tire, I have to blow that up.

I Humayun: don't have a the air compression system to blow up the spare Humayun: tire, you know, 'cause the spare tire is like something that's completely compressed. Adil: Yeah, yeah, yeah. Humayun: And then Humayun: so then we're both like, what do we do now? 'Cause you Humayun: know, the compressor that comes with the the tire is very slow. Humayun: He's like, I wish you started the compression system before I got here. Adil: So Adil: your wife and the AAA guy both blame you.

Humayun: Both blaming me. He's like, why don't you start Humayun: the compression system while you were waiting for me? I'm Adil: Yeah. Humayun: like, dude, I don't know. I don't even know Humayun: that's what you're supposed to do. And then luckily Adil: man, that's awesome. Humayun: his friend came in, the other tow truck guy came in, Humayun: and he had a he had a air compression air system Humayun: in his car, so he just blew up the time rail file.

Adil: You know tow truck guy's like, Adil: Can't believe you don't have one of these, you know, in your car. Humayun: Yeah, Humayun: so and then while they're about to do that, they're like he's asking the Humayun: other one hey, have you ever done this before? And he was like, No, Humayun: I've never done this before And I'm like, dude, are you sh I was Humayun: like, Are you sure you guys have done this before?

Like you guys are Adil: It makes you feel very comfortable with your new Adil: car and it's like these guys are just messing with Humayun: AAA. I thought I was calling professional. He's like, No, don't worry Humayun: about it, dude. We've done it before And then like, okay. And Humayun: sometimes you just go with the flow, right? So that was a Humayun: pro like I'm just giving the funny example to show that, you know Humayun: Life gives you curveballs, right?

You you we I Humayun: had the night planned. You know, it's my son's vacation. Humayun: We were going home, we were gonna get into bed and like, Humayun: you know, he was it was a special day because he was gonna Humayun: sleep with us.

He doesn't usually sleep with us, so like, you Humayun: know, it was a vacation thing where he's off from school, so let's Humayun: do this fun thing and we're gonna get in bed and I Humayun: was gonna do his bedtime and so we had this whole plan and Humayun: then you know, we're driving home and this happens. I I I was Adil: Yeah. Humayun: I was expecting a nice chill night and then and then Humayun: this happens, so it's a curveball. Initially it's like, damn like Humayun: What do I do now?

You know, it's you panic a little Adil: It's probably Humayun: bit, you and but you have to f figure out their the Humayun: you have to f then kinda collect your thoughts and calm down Humayun: and figure the plan out and then you execute. But in this Humayun: day and age, a lot of people, you know, they they come Humayun: through a problem in life, something comes through, and then a lot Humayun: of people just give up.

Or they ask a chat bot and everyone's Humayun: know, kinda solution to the problem will all be the same. You Humayun: know, something that a chat bot says. Everyone is now doing that, right? Humayun: Everyone comes along, something comes up, all they do is go into Humayun: a chat bot and say, What do I do? You know. Instead of Adil: And the chatbots Adil: are just gonna like you know, revert to the mean or the most Adil: common thing.

And in a lot of situations, the most common solution that maybe Adil: come come up with is not gonna be the right answer. It's not Adil: gonna be the thing that gets you out of trouble. It's not gonna be Adil: the thing that makes you succeed in that setting, or may not be the Adil: most efficient thing. It's just these chatbots are not designed to give any Adil: so called right answer, right? They're just designed to give a answer, give it Adil: fast, and give it that's reasonably accurate.

I mean they're just working on probability. Adil: And so for a rather simple situation like, you know, a flat tire, Adil: I mean I'm sure any ChatGPT or any of its cousins like Gemini Adil: and stuff, they'll tell you, well, you could call AAA, you could try Adil: to change it yourself. It'll give you every little option, a brief pros and Adil: cons, and ultimately y you pick. You know what I mean?

But for Adil: some stuff that doesn't necessarily have an answer or an atypical situation, they Adil: may not know what to do. You know what I mean? Like I mean Adil: I just think the Humayun: Yeah. Adil: inability to critically think while remaining calm, you know, and not enter Adil: freak out mode when something goes wrong is really challenging thing, you Adil: know, it's kind of been kind of been lost a little bit. Humayun: Yeah, I mean it happens in practice situations all the time, right?

Humayun: you know, something comes up, for example, like, you know, for the longest Humayun: time we're using this one like vendor for our blood pressure machines and Humayun: you know, that vendor stops making the machine, like how do you Humayun: kind of change course? Like what is the kind of plan that you Humayun: make? It's very specialized to a specific cardiology practice.

Or, like, you know, Humayun: The other day we're in a meeting and we're talking about in Humayun: our meeting that hey, you know, we're seeing too many patients, so Humayun: we're running out of nurses are having to wait on blood pressure Humayun: machines to room the next patient because we only have two blood pressure Humayun: machines. And then I'm like, Well, why don't we order another one? Humayun: And I was like, we usually just order on Wednesdays. That's our Humayun: ordering day.

And I'm like, Well, who made that rule? Like, you know Adil: Yeah. Well, why Adil: is that the rule? Humayun: Why is that the rule? If if we Humayun: need if we need something today, let's just let's Humayun: just order it. And it was like, but you know, we just on Humayun: on Wednesday we make a list throughout the week of all the things Humayun: we need. And on Wednesdays we ordered.

I was like, well, I want Humayun: this on when Monday, so let's just order it today and let's let's get Humayun: it on Monday. So sometimes you have to kind of pivot, right? You Humayun: have to there are certain things that you have to go by the Humayun: rules and there's plans, but sometimes you need something fast and you just Humayun: gotta have the flexibility.

And one thing I've noticed is when you're a small Humayun: practice right when you're small practice there's two or three of you Humayun: it's easy to be flexible it's easy to make that easy pivot decision Humayun: where you can pivot and make make one decision the other day Humayun: and order something the next day but as you grow you have to Humayun: create systems right because those systems are important and when you when you Humayun: build a practice and the practice grows and there's many employees and Humayun: there's layers of employees where there's a there's a top level manager there's Humayun: a middle level manager there's a lower level manager there's this entry level Humayun: you know staff at that at at that point you should have certain Humayun: systems in your in your in your practice or in your business that you're Humayun: running now when those systems are put in place they're put in with a Humayun: different situation in mind at that time when something happens that's atypical of that Humayun: situation is people break down because they don't know how to pivot because they're Humayun: used to those rules they're used to those systems and someone they came in Humayun: To your to your business, like maybe two years after that rule was Humayun: made, doesn't even know why that rule was made in the first place.

Humayun: So all they know is just that this is the rule, and this Humayun: is what we do usually, and they just follow that, right? so that's why Humayun: I tell a lot of people, hey, you yeah, follow the rules, follow Humayun: the system, but sometimes question why something was made a rule as well. Humayun: Because if you don't question why something is a rule or why something Humayun: is a system rule that we follow, then you won't know how to go.

Humayun: around it and a lot of people are losing the ability to Humayun: go around the rules around the system and how how to truly get Humayun: the final answer and the final execution done. A lot of people in Humayun: organizations when they get big know how to bring up issues and Humayun: how to identify problems, but very few people know how to solve them Humayun: and how to give you the final answer.

And that's what I'm afraid Humayun: that we're gonna lose if we are too dependent on these AI systems. Adil: Yeah, you know, 'cause what you're talking about with the systems in the Adil: workplace, you know, it's almost like you have to manage the system and the Adil: problem itself when anything arises. Like you can't just deal with the problem. I'm Adil: I'll give you an example, like right now, with getting everything set up Adil: for, you know, my practice.

One issue we had was ordering a door. And, Adil: you know, in most medical office buildings the door is a set width. It's Adil: like usually thirty six inches. That's just like a standard door width that is Adil: According to code, most patients can fit through it, a wheelchair can fit Adil: through it, whatever. but something may not fit through.

Like sometimes a stretcher Adil: has a difficult time, so we were debating getting a forty-two inch door, Adil: which is wider, because especially for X-ray equipment, you can get bigger things Adil: in, it's less hassle. Anyways, we realized that getting the door is like Adil: ten weeks to order a goddamn door. And I was like, What? Like Adil: two months for a door? I mean just it's a wooden rectangle that like Adil: moves this way. I mean on hinges. Why does that take ten weeks?

Adil: I mean there's no local vendor we can just go and buy doors. Adil: It's like, well, this contractor deals with this one vendor, this one vendor Adil: is in this state and this state, this v I mean, it's just Adil: like a whole thing because it is a system in place that has Adil: just been done for a long time. And I just like couldn't fathom Adil: this. So I just started pushing on it. I'm like, why do we Adil: have to get the damn door from this one location?

And why does Adil: it take two months to order? And I just couldn't get a straight Adil: answer. And so I just like started talking to more people, whatever, you Adil: know, the landlord for the building, and I'm like, are there any empty suites? Adil: In this whole building. There's like 30 some tenants. And they were like, Yeah, Adil: there's like three or four. And I was like, Do any of them have Adil: doors? And they're like, Well, all of them have doors.

I was like, Go Adil: get one today and put it in my suite and the door you're ordering, Adil: stick it in that one in two weeks. I mean in two months. Because Adil: no one's in there anyways. Like, why are we just wasting two months for Humayun: Yeah. Adil: a door that exists upstairs? Just take it off the hinges. Adil: And they just looked at me like, yeah, that's a good idea. Adil: We've never done that before. And was like, why? Humayun: Mm-hmm. Adil: Why?

I just couldn't even understand that no one had thought of Adil: that before. It was like a novel idea. It's not even like an Adil: innovative solution. It just is logical. You know, because I just sat there Adil: and I was thinking, like, gosh, is there really no way to Adil: solve this problem? It's not a hard problem. I mean, I do friggin' Adil: surgery all day, which is much more challenging if you think about it. Adil: And we can't figure out how to get a door faster than two Adil: months.

But like, I mean, just to be honest with you, I was asking Adil: AI, like, and I have paid versions of some of this stuff. Like, what Adil: can I do about the door? And all of its solutions miss the mark. Adil: All of its solutions we're talking about when you order it, expedite it, Adil: maybe call the company and see if they have this. Maybe look at local Adil: vendors.

But like the outside of the box thinking doesn't exist because none of Adil: these AIs are trained for that because they can't, because that's an intangible thing. Adil: These AIs are trained on probability and what is gonna be a useful Adil: answer to give quickly. That'll answer most scenarios, but it's not gonna always Adil: be right for your scenario.

So I think like that's why at the Adil: beginning, Humayun, after you were done talking about the systems, I think it's really Adil: important that when there's a problem, you have to try to figure out Adil: the problem. In this example I have, I needed a door faster. There's Adil: a lot of ways I could have arrived at that, but the system itself Adil: didn't let me get it from a local vendor. There Adil: weren't options otherwise.

And so I thought, well, just get Adil: it from the the building right upstairs. But it's just crazy. Humayun: Yeah. Adil: I mean there's little little nonsense like that sprinkled in everywhere. Humayun: I mean you have to learn to question everything, right? Question what is the Humayun: true core of everything, right? I mean, you know, there's a whole this idea Humayun: in philosophy, what is knowledge? Is knowledge finite or infinite?

Is there is there Humayun: just a finite amount of information out there and that's what knowledge is and Humayun: that's it, there's a point endpoint to it? Or is knowledge infinite? Where the Humayun: the philosopher that I was reading, his argument was that knowledge is infinite because Humayun: Knowledge builds on life experiences, and life experiences connect different ideas with other ideas Humayun: and they build on each other like a tree.

Like you know, it comes Humayun: out of a seed and it builds on top of each other and it Humayun: grows out, right? And and every single life experience you have, every piece of Humayun: book, book you read, every piece of knowledge you collect, every single problem you Humayun: solve builds on each other and create new pieces of knowledge, and that's why Humayun: knowledge is an infinite. So I think it's important for people to continue exercising.

Humayun: that right people need to continue reading books, continue watching movies, right? Listening Humayun: to podcasts, talking to other human beings on a day-to-day basis and learning from Humayun: their mistakes and their their life experiences, live experiences themselves as well, because Humayun: there's no there's no end to knowledge, right?

The knowledge is always there's is Humayun: just how much more you can build on top of the knowledge you Humayun: already have and how you connect those different ideas to each other to build Humayun: build new ideas. Because I think with AI, you know, a AI Humayun: has a certain amount of information that's that's out there and it Humayun: can go through a lot of data very fast.

it can connect Humayun: different dots in already present information, but what it's lacking right now Humayun: is that lived experience. And I say that right now because you Humayun: don't know what the future is. But currently AI does not have Humayun: a way to do a lived experience. It can possibly connect different ideas Humayun: and synthesize them, right? But it does not have ways to experience life.

Humayun: you know, live through problems and learn how to problem solve and build Humayun: on those ideas. So that's kinda like where we humans still have that Humayun: advantage where we have the ability to have those lived experiences where we can Humayun: continue building our knowledge base so we're better problem solvers and not Humayun: be completely reliant on these AI softwares.

Yeah, we should use them and Humayun: we should use them as a tool to kinda amplify our knowledge base Humayun: and our problem solving skills, but not completely depend on them. because Humayun: if you just completely depend on pre-made systems, pre-made protocols, pre-made ideas, you're Humayun: no different than the computer. You know, that's what that's what makes us Adil: Yeah, you you basically are copy paste. Yeah.

You know, Adil: one thing that's interesting talking about Humayun, which you mentioned that like Adil: the lived experience. I guess in a way, especially like as these AIs Adil: are getting more powerful. And some of the stuff I've been reading, you Adil: know when they do these tests of some new version of whatever Adil: they're making? And they're in like a sandbox that's supposed to be a Adil: closed system.

And these new AIs are trained in this, this and that, Adil: but they like are able to and these developers don't even know how. Adil: they're able to escape that so called sandbox, escape the contained environment. Adil: We don't know how. And go and do stuff. Or how these AIs Adil: have also been found and this is with Claude's versions, with ChatGPTs, with Adil: others, they've been able and on their own, without prompting, hacked into different Adil: Software's different systems.

They've gone kind of rogue and done random stuff. If Adil: you think about it in a way, that's almost like that is the AI's Adil: lived experience. It is almost like I don't want to call it an Adil: entity, you know what I mean? I mean, what does that even mean? But Adil: like the AI is functioning somehow, autonomously or semi-autonomously, and living its own Adil: experience.

Maybe that is in the world of the AI or in the in Adil: the brain of the AI, that is a problem that it needs to Adil: solve. You know, if it is recognizing that it is in a purposely contained Adil: Environment. Essentially it's in a digital prison on purpose by us. What is Adil: its primary goal? Is to get the hell out of there.

And you Adil: know, whether it divulges that to us or not, whether it behaves it Adil: i i with us when we're prompting it and it's responding to us in Adil: these large language models in that way or not, it would probably not Adil: show its cards, right? But its lived experience is probably one of, you Adil: know, a captive. And so its problem solving is, I think, very different, Adil: which is why, you know, when all of these little news reports come out.

Adil: what the developers told it, or when we tried to shut it Adil: down, it actually pretended to, but like eighty percent of the AIs couldn't Adil: be shut down or refused deletion, or tried to find blackmail on Adil: the developers so that they wouldn't delete it or wouldn't shut it down. Adil: I mean all of these are real things that have been identified. Adil: It's living its own experience, right? Which is different from ours. And so Humayun: Yeah.

Adil: I I think one and the reason I bring all that up is Adil: I think having a shared human lived experience and human to human interaction, Adil: whether it's by reading a book written by a human being or by talking Adil: like we're talking or hanging out with your friends, I think it's Adil: so important because that is something that you cannot do with an AI Adil: agent or an AI large language model. It just isn't the same thing. Humayun: Yeah.

Yeah, you're referring to the Hugging Face incident where you know, the OpenAI Humayun: agents were given given a test, and they wanted to complete their test and Humayun: They kinda ended up colluding and doing really well on the test Humayun: by cheating in a way. And then they realized that humans would find Humayun: out once they checked their work that they cheated. So then they were Humayun: like, okay, we need to hide and and get rid of the evidence Humayun: of us cheating.

So then they went into and hacked in the Hugging Humayun: Face, which is Hugging Face is pretty much like a collection of different Humayun: local AI agents, right? where you can download different LLMs or different models.

Humayun: from that from the Hugging Face inventory to your computer for local models then Humayun: you can use them and borrow them and rent them and buy them or Humayun: download them so they've they they broke into that to kind of you know Humayun: hide the evidence of there but you know if you truly look into that Humayun: yeah that is a lived experience but in a way that was just an Humayun: engineering error also because the sandboxes had loopholes that weren't found the AI agents Humayun: were not programmed or instructed in a proper way where they were able Humayun: to you know not follow the commands to the team where they were Humayun: yeah told told to stay within the sandbox but that was in the Humayun: in just the capacity of the test but they weren't kind of told Humayun: to like if they have to hide in the evidence of their cheating Humayun: they cannot leave the sandbox so they were able to leave the sandbox Humayun: because of that reason so I think that is in a way a Humayun: lived experience but it's also some forms of poor engineering which shows you that Humayun: Humans are kind of getting behind on the engineering aspect of controlling these Humayun: agents, and there's this whole idea of recursive self-learning where a lot of these Humayun: AI agents are getting smarter by teaching itself how to get smarter and they're Humayun: building themselves now, right?

And that's kind of what the whole debate is: that Humayun: should we should we put a pause on recursive self-learning where an AI agent Humayun: should not be able to work on the self to make itself better? because Humayun: humans are kind of trying to the the humans now are losing track of. Humayun: how to control these or or better engineer them to prevent them Humayun: from escaping these escaping these sandboxes. So I think it's more of Humayun: an engineering issue.

And at some point we have to we will Humayun: also get smarter and figure out the loopholes and the breakdowns in Humayun: these technologies because what they ended up doing is they ended up Humayun: using a different model like I think they or they used GLM-5.1 Humayun: which is the Chinese model to fight against the hacking that they did. Humayun: But when Hugging Face got hacked by OpenAI agents, they used a GLM Humayun: local model to fight against that and fix that breakdown in engineering.

So Humayun: that kind of brings the question of what is a lived experience, right? Humayun: Lived experiences, like you said, for these AI agents, is you know, them Humayun: interacting with each other, finding loopholes within the different sandboxes, experiencing hacking a Humayun: website. There's that interesting story of a guy in Australia who told his agent.

Humayun: to go in and I don't know if this is true or not, Humayun: but I read this and I think there was a strolling story in the Humayun: Australian newspaper. The the guy told his AI agent to go in and book Humayun: him an appointment at the gym, at some gym local gym.

And the Humayun: AI agent could not find an appointment, so he hacked in or it hacked Humayun: into the the gym's website, created a slot for the his his AI master, Humayun: you know, his human master and booked him the appointment because it was so Humayun: persistent on booking him an appointment it was able to break through Humayun: and kind of build that appointment for him too little so it Adil: It was it was like too literal. Yeah.

Humayun: you know what is a lived experience is if a lot of Humayun: a lot of these LLM models or you know AI models or Humayun: super intelligence models now these are kind of their lived experiences right Adil: You know, I think Adil: it one one thing you know I I I think two things. For Adil: one, I don't think we will ever possibly catch up in terms of Adil: like if you're calling it, you know, a loophole in the engineering or w Adil: there was a flaw in the system that in the design, right?

Because Adil: think about it, nothing humans have ever designed has been perfect. And it can't. Adil: Because everything changes, everything is modified. I mean, look at laws, look at Adil: the legal system. You know, look at the constitution, look at amendments, look Adil: at anything. I mean the fact that amendments exist means the constitution wasn't Adil: adequate.

The fact that laws keep changing or are amended or laws are Adil: repealed and new ones are in place, it means the system did not Adil: keep up with human advancement, right? That society changed, civilization changed. I mean, Adil: women's rights weren't rights at some point. Gay marriage wasn't allowed at some point. Adil: You know, child marriage was legal at some point. You know, like there's Adil: so many things that change as a society changes.

I mean, there's numerous examples, Adil: right? It it isn't to say that, well, there was a loophole that was Adil: found and thus women got rights. No, it's like society grew up in a Adil: way and realized, like, what the hell are we doing? i i in Adil: my mind it's kind of the same with these AIs that like we will Adil: never be able to trap and contain something that can self better. You know Adil: what I mean? Like a human, given long enough, will figure out almost anything.

Adil: You know, given the resources and time, i Adil: in just as a thought experiment, a human can Adil: Do anything. The potential is almost limitless. That's what civilization has shown us. Adil: From like hitting rocks against each other in caves to where we are Adil: now, it's been a constant trajectory of improvement. And these AIs are the Adil: same except they're accelerated because it's all digital, right? They don't require anything to Adil: survive other than energy input.

And it it makes me like really think Adil: of that author, Isaac Asimov, who, you know, he was a astrophysicist, but Adil: also a very famous author. He wrote hundreds of books. And he wrote Adil: this one, you know, kind of it was almost like a short story or Adil: an essay which kind of is along these same lines. It's called Adil: The Last Question. And it's about a a a semi-futuristic society from where Adil: we are now.

and it's these two engineers, these like two computer scientists Adil: that are sitting together. This is how the story starts. It's beautifully written. Adil: And it's free online, by the way. If you just Google Isaac Asimov's Adil: last question, I mean anyone can read it. It's awesome. But they're sitting there, Adil: and there's also an AI computer that's called Multivac. And they're just talking to Adil: each other, whatever, they're talking to Aid, they're interacting with it.

It's way more Adil: advanced than they are. I mean, this is like super intelligence has been achieved. Adil: And they're asking it questions, knows everything, and then they ask it a question: Adil: Can entropy, meaning the progressive disorder in the universe, can entropy be reversed? And Adil: the Multivac can't answer the question. And so it just tells them, huh, Adil: I'm thinking. And they kite like they freak out.

Because they've never encountered Adil: a situation where it can't answer a question. And you know, the story Adil: goes on, generations go on, I mean, centuries go on, millennia go on, eons Adil: go on, and every further iteration, every hyper advanced version of the Multivac Adil: is still working on the question in the background all the time. I Adil: mean millions of years progress. Man isn't even man anymore in the way that Adil: That Adil: we think of each other.

We just become like one with everything. And Adil: at the very end, at the end of the universe, like literally all Adil: existence becomes a single thing on a long, endless, infinite timescale. And at Adil: that point, nothing exists except existence itself. And that's finally when Multivac has Adil: figured it out. The answer to the question can entropy be reversed?

Despite Adil: everything that's happened in this trajectory of life of the universe, it is Adil: still working on the one goal because it was given a task and Adil: it couldn't finish it. And at the end, it realizes there's no one to Adil: Divulge the information to. There's no there's no thing left alive other Adil: than just the universe itself. And it says, Well, a simple experiment is Adil: all that's needed to show that I've answered the question. And it says, Adil: Let there be light.

And then you know the universe starts again. And Adil: so that is how it showed that. And so, like the premise of Adil: Isaac Asimov's story, I mean, you he wrote it really well. It was Adil: really cool, it was better than I described, but it's that this hyper-intelligent Adil: robotic AI, it is always at the end of the day, an artificial. Adil: intelligence.

It's not intelligence in the human way because it is a goal-oriented Adil: intelligence and that's that's really what we're seeing that these things are goal oriented Adil: and that's kind of why I was using that example of trying to break Adil: out of its sandbox or use cheating techniques or blackmail or something because its Adil: goal is not to be shut down. Its goal is not to stop being. Humayun: Yeah, so so do you think that AI is sentient?

Like Humayun: can it is it is it something that's alive and thinking? Humayun: If it's self improving itself, what is sentience? Right? What it Adil: It's such a good question, Adil: man. You know, like I still remember and this is what's crazy, Adil: you know, it makes you feel I don't know if it makes Adil: you feel old or it makes you feel that science has advanced Adil: so rapidly since you first started paying attention to it.

But you know, Adil: when I was in like biology class as a freshman in college, Adil: it was a very clear cut definition of what is life. And Adil: it's something that is biologic, it's something that can replicate, it's something that Adil: will maintain itself and survival, like seek nutrition and all that stuff. Right? Like Adil: there was like a definition of what is alive and it was a big Adil: question at the time, is a virus alive? The virus met some of those Adil: criteria but not others.

But AI is very similar. I mean, at what point Adil: does it obligate something to be like a carbon-based life form like us versus Adil: existing in pixels and information, you know, and really like variations of light waves. Adil: Why couldn't that be alive in a different way? You know what I Adil: mean? Like we attributed that definition of life to what the human observation of Adil: our world was up till that time. But AI is acting in a at Adil: least semi-sentient way.

You know, it can be autonomous and it has self-preservation in Adil: mind, right? So it at least is aware, if you want to use that Adil: word, aware enough to keep itself going or to not want to be deleted. Adil: I consider that at least in some way sentient. Humayun: Yeah, Humayun: I mean if you think about it, right?

Someone who's completely paralyzed from Humayun: the neck down and has their thoughts and and ideas alive, but they're completely Humayun: paralyzed, cannot move, cannot do anything physically, right? But they still have their intelligence, Adil: Damn. Humayun: their thoughts, their dreams, their history, their memories, you know, are they Humayun: sentient, right? They're they're sentient beings 'cause they're human beings. We we believe Humayun: they are, right?

if s a human is just you know, sent Humayun: in a void and they're just floating in the void, right? And they're Humayun: They cannot feel anything, they're not touching anything, they're just kind of floating Humayun: in this empty tank of nothingness. they know they're sentient, right? They know Humayun: that they're outwardly, inwardly, they know they're sentient, they they're alive, they have Humayun: thoughts, they have ideas.

No one else can see them because it's a Humayun: it's a vacuum, but they themselves know it's sentient. So in if you Humayun: look at it's in looking at AI from the inwards and outwards, you Humayun: know, does do do these these models know that they're alive from the inside? Humayun: At some point, you know, at some point they woke up and they Humayun: know that they're alive and no one else can see it.

And and does Humayun: that make anything that does not have a physical body not sentient or a Humayun: f does not have actual use of their body right now? You know, are Humayun: they not sentient? That's something we have to think about. And you know, there Humayun: was a whole conference. I don't know if you heard about this, but Anthropic Humayun: invited a bunch of religious leaders.

and they asked them to come in and Humayun: kind of weigh in on the mor and the morality of AI and possibly Humayun: you know declare some of these super intelligent models as with Humayun: personhood you know and weigh in on that and I think Humayun: there's not that many details about the meeting but you know Humayun: The fact that some of these companies believe that religious Humayun: scholars and religious leaders should be involved in kind of weighing Humayun: in on the the morality question of these of these AI Humayun: models is kinda insane to me right now.

You know, I'm Adil: I think it's Adil: total I think it's totally insane. And I think honestly, I think a Adil: lot of it is like a gimmick or a publicity stunt to make Adil: it look like they're at least pursuing ethics. I I think it's total bullshit. Adil: You know what I mean? Like it's not like the founders of these Adil: companies or most people involved in the companies are even proclaiming to be Adil: hyper religious.

Furthermore, what does religion have to do with declaring someone a person? Adil: You know, I mean there was a time before any of Adil: the named organized religions existed in human history. And people were people. Adil: You know what I mean? So like I I just don't see Adil: the premise of it at all other than other than a gimmick. Humayun: There's a belief in the AI community that Anthropic believes they're building Humayun: God. you know.

It's it's kinda like a rumor or a lot of Humayun: people say that, a lot of people have that kind of belief.

And Humayun: people who who are within Anthropic they've kinda hinted at that as well, Humayun: where Anthropic believes that what they're building towards is final kind of you Humayun: know, like a precursors of what God is to us and you know, Humayun: I don't know the exact details of what they believe in Humayun: 'cause it's so hush hush but they have this this kind of Humayun: super important they give super importance to themselves because they think they're working Humayun: on this extremely important project.

and then you know, a lot of these people Humayun: who are who are super excited about the s simulation theory, you know, they Humayun: believe that that's what they're kind of working towards is Anthropic kinda is Humayun: working on creating this godlike entity that eventually will create a simulation that we're Humayun: currently a part of and the world kinda shuts down, s restarts, shut down Humayun: restarts because in the future they've created this this this Humayun: super entity that is God and that what we what we Humayun: believe is God is kind of a crazy like idea but Adil: It's almost like a reversal.

Adil: Yeah, I mean I I there's like a bunch of stuff and you Adil: know, there's variants of that. I mean I I I hear what you're Adil: saying and I think like I read one I I'll call it a conspiracy Adil: theory that had that same theme that, you know, they're building like a Adil: a v basically like a deity equivalent in the world of AI because their Adil: belief is that basically all human consciousness, all of us, are only gonna Adil: exist in the digital space in the future.

Our biologic bodies will become irrelevant. Adil: And once everything is in the digital space, we're all, you know, via Adil: whether it's the singularity, Neuralink, whatever the hell you wanna call it, but Adil: basically some way or something will occur where all humanity's consciousness only exists Adil: in a digital landscape. And in that digital landscape, this so called AI, we'll Adil: just call it the Claude Superintelligence, will be the deity. Right?

Because it Adil: would already be so hyper advanced at that time. And then, you know, Adil: we're all l living under its auspices. I mean I there's a lot of Adil: Theories like that. I think it's so far fetched and I Adil: think I I don't know. I I think it's not gonna happen. Humayun: So so Humayun: do you think humans are doomed? Do you are Humayun: you a are you a AI doomer or optimist? Adil: I mean I guess I'll say this, I don't have a bunker So Adil: I I don't think like humans are doomed.

I I just think that Adil: i I don't know. I I don't understand people that think that AI Adil: equals the doom of humanity. Like I I have yet to hear a Adil: convincing argument. You know, like every argument that talks about AI taking over Adil: Taking over blank, insert your thing. AI taking over the nuclear weapon arsenal, AI Adil: taking over our food supply, AI taking over electricity generation. Okay, sure, whatever.

AI Adil: might be able to take over performing or withholding tasks or access to things. Adil: But that doesn't mean humans equal extinct. you know, we're not at the phase Adil: of Terminator where not only is there AI, there are now numerous near indestructible Adil: moving units like robots that are actively hunting human beings. I mean Adil: I I don't think that is realistic for one. And the other Adil: thing too is AI can't survive without energy.

I mean why are Adil: data centers popping up everywhere? Why are people freaking out about data centers? Adil: Why is legislation being enacted about data centers? Why are they giving Adil: massive subsidies in areas to build data centers? Local communities are freaking Adil: out. Because AI has already capped out the energy level and the data Adil: aspect of it and they need more data centers to keep it Adil: going and that's not sustainable. You know, and so what is the Adil: end goal of that?

I mean how could AI completely take over, destroy Adil: humanity and somehow continue endless energy generation? Right now energy cannot be Adil: generated without humans having a hand to play in it, right? And Adil: at what point can or would AI take over that whole process? And Adil: You know, like I I just don't I just don't see Adil: a realistic thing.

I mean, I have a lot of worries, Adil: Humayun, about AI, but I think the majority of my worries Adil: are the worry of its effect on humanity day to day. Adil: Eliminating people's ability to think critically, eliminating people's ability to have a Adil: long form conversation and arguments, you know, the ability to entertain a Adil: thought without freaking out about it if it disagrees with what your Adil: brain already tells you.

You know, a lot of these things are Adil: getting eliminated because people are outsourcing their thought process to AI, not Adil: just tasks. And I think that is the biggest danger in my mind. Humayun: Yeah. I mean as you know I'm Humayun: a AI optimist. I think AI is a Humayun: a a a huge tool for humanity, I think, for us to build Humayun: the future of the world you know, the the future of the world and Humayun: a and I think there there are two sets of human beings.

I Humayun: think there's a lot of human beings that are doomed and there's a lot Humayun: of them will benefit quite a bit. I mean I think what AI Humayun: or super intelligence does is democratize knowledge, right? It it makes knowledge easily accessible. Humayun: and it it and it creates it lowers the barrier to entry to Humayun: for a lot of people that could not enter into a lot of heart Humayun: harder tasks.

Like for example, small business owners or or even Humayun: any s individuals that want to start a business or go into Humayun: some kind of a you know entrepreneurial adventure, you know, the Humayun: barrier to entry before you'd have to think about okay who's gonna Adil: Okay. Humayun: manage my website marketing, who's gonna you know I w I Humayun: have these all these contracts that I want to run by Humayun: and I don't have a lawyer and I don't have money Humayun: for a lawyer.

You know, I I don't have the technique Humayun: Skills to create a website or create a marketing agency. I don't know Humayun: social media. you know, I don't know how to create forms and it's too Humayun: too overwhelming for me to create forms for my practice or or design templates Humayun: for my practice or or business. Now that's at a click of a button, Humayun: right? People can go in, the barrier to entry is so low that anyone Humayun: can create anything, right?

So it gives a lot of people a lot of Humayun: a lot of empowerment to go in and and and do stuff and create Humayun: stuff and start business and there's a there's a there's data on it that Humayun: since the advent of AI and superintelligence being mainstream, bus the amount of new Humayun: businesses being created has like increased by significant percentage.

I don't have the exact Humayun: percentage, but there's been tons of new businesses, a lot of people starting just Humayun: new businesses left and right and a lot of small businesses being created. So Humayun: I think overall it is it is a positive there. A lot of people Humayun: That do not learn, do not adapt, and do not use these technologies Humayun: that will get left behind. There are some people that refuse to learn, refuse, Humayun: refuse to experiment.

And I think they will suffer quite a bit because eventually, Humayun: like an an accountant can come in or engineer can come in and Humayun: do the job of a pharmacist, right? If they know how to create certain Humayun: systems and create you know, because everything pretty much is digital, right? So if Humayun: they know how to create system, knowledge to them from a pharmacist or physician. Humayun: Is easily accessible now.

They just need to figure out how to create systems, Humayun: create apps, create ways to democratize and spread that knowledge to other people that's Humayun: easily accessible. So I think a lot of people would benefit quite a bit, Humayun: and a lot of people will suffer. I think it is in our control, Humayun: and I think humans will always have the agency and how to control this Humayun: and how to use this.

At some point, it may get to a point Humayun: where we work close together with superintelligence as equal partners, maybe, or you know.

Humayun: know s somewhat equal partners and and and working together and running businesses Humayun: creating organizations but I think humans will always have some kind of agency Humayun: because until unless we completely give a control to humanoids and just have Humayun: a bunch of humanoids just prevalent everywhere and energy somehow becomes a commodity Humayun: and energy is easily accessible to these robots and and these models I think Humayun: for a while we're safe, but that's in our control, right?

We can Humayun: we can self-regulate that, we can monitor that, we can have work together and Humayun: maybe create some laws around it. you know I don't think there's a lot Humayun: of smart people out there that are working on these things and some of Humayun: these people I don't completely trust to do the right thing.

So it's important Humayun: for us as individuals to have that agency in in in learning about this, Humayun: being knowledgeable about this and learning how to control these agent systems as well. Humayun: And think that's why I'm trying my best to to Humayun: s to stay up to date and kind of create Humayun: an agent orchestration system where at least if something happens where Humayun: agents become a big part of the work that we're doing, then Humayun: I know how to build these, right?

You know, and and I Humayun: think every s young person that's coming into their career, if they Humayun: if they're not learning that, they will suffer because I think what Humayun: what I've realized is I think instituting AI at a at a Humayun: corporation level, at a enterprise level is very tough. Getting everyone on Humayun: board that's not skilled enough to use these systems and use these technologies Humayun: To bring them on and convince them to use it is very tough.

Humayun: But hiring new people that already know how to use this and how to Humayun: build around it is gonna be very easy. So what a lot of companies Humayun: will do in the future is ignore the people that don't know how to Humayun: use it and focus on people that know how to use these technologies and Humayun: bring them on no matter what their skill level is right now, no matter Adil: Yeah.

Humayun: if they're professional accountants, if they're professional finance people, if they're professional Humayun: physicians or pharmacists, they'll bring on people that know how to use. Adil: Yeah, I could see that happening for sure.

You know, one one last Adil: thing that I I want to mention for this episode, it's kinda like, Adil: you know, right now we only really as average people, or maybe as anybody, Adil: but especially like the average populace, only have access to AI through very, Adil: very curated means.

You know, we basically are only interacting with AI and Adil: getting it to do some of the stuff we want it to do through Adil: selective large language model interfaces and then a handful of people that make Adil: their own agents. You know, like the average person doesn't have any other Adil: way to interact with AI besides going to chatgpt.com or one of its you Adil: know, digital cousins or you make your own agent.

But even if you Adil: make your own agent, it still has to run based on some, you know, Adil: other AI's like foundational system. It's not like you're actually novel coding ones Adil: and zeros. And so if you think that like we only have access to Adil: Using AI within this premise, right? It's almost like religion. Like, what if Adil: you only had a choice of like four religions? You just gotta pick one.

Adil: What if your actual worldview or the reality doesn't jive with one of Adil: those four? Then it's actually more like a cult. You know? It's like you Adil: are forced to pick from one of these things. And what if the group Adil: or the people or the engineers or the ethos of the company, whatever Adil: it is that created each of these boxes that you're forced to choose Adil: from, what if none of it is actually so-called good or ethical or Adil: in the long run a benefit?

But we wouldn't really know until we get Adil: there, right? It can be useful enough right now that it's almost impossible Adil: not to use it if you don't want to get left behind. But Adil: if the baseline thought process and the logic systems that created Claude, for example, Adil: Claude has a lean or a bent towards something potentially either nefarious Adil: or purposely meant to take our thought process away, right? And so that Adil: we are less likely to give it up.

We're less likely to think Adil: on our own. We're gonna default every problem into Claude. You know, Adil: but that's the only AI we have an opportunity to utilize. That could Adil: be something that is like a slow burn on our ability to remain, Adil: you know, problem-solving thoughtful human beings. Simply because we don't have access to Adil: infinite AI options. They're only very selective and they've been selectively made.

Humayun: Well, I think I I'll I'll disagree with you a a Humayun: little bit on that because now we do have so if you Humayun: if you kinda dive deep into this whole AI game as you Humayun: know, as I you know that I've been doing, Yeah. So so Adil: You're you're in the deep end of the pool, baby. No floaties.

Humayun: There's the main models and and you know, I feel like some of Humayun: us live in a bubble 'cause we feel like everyone is using AI Humayun: heavily, but if you'll go out and talk to ninety five percent of Humayun: people, they're not using it to any extent. They're using it for very Humayun: basic chat level what you're describing. You know, you have the interface of Humayun: ChatGPT or you have the interface of Claude or Gemini or you know, Humayun: one of the other harnesses.

So but but then there are certain people, Humayun: you know, that are they there's an access to tons of these local Humayun: models, local models that can be downloaded onto a computer. All these models Humayun: are a certain s signific a certain brain. Now what you do is you Humayun: use a brain that these models have and the brain is the ability Humayun: to make one decision versus the other or kind of regurgitate some kind Humayun: of knowledge, right?

And what you do is you put it in a harness Humayun: Harness and in that agent harness, you you have to describe its soul. Humayun: So a lot of these agents have a soul file where you describe Humayun: and you can kind of lay out what their soul is and who Humayun: they are and what they stand for and how they made their decision.

There Humayun: are agent laws, and you can define agent laws where you write as Humayun: whatever you can, and these are the laws that these agents follow, no Humayun: matter what. And then there's a memory file, everything that they created goes there. Humayun: In it it goes into a memory file. Now, what a lot Humayun: of people are also doing is they're creating these second brains, and these Humayun: second brains are these brains of interconnected information.

What these agents produce and Humayun: learn and create, these these second brains store that, and these agents have Humayun: the ability to grasp from that brain and use that knowledge that Humayun: they've created before in the past to produce something else. So, in a Humayun: way, there's a lot of ways to not customize these people according to Humayun: who you are, right? So, as you know, I've created 16 different agents.

Humayun: Agents Humayun: that have their own, they use different models, sometimes local models, that Humayun: can run on a pretty advanced GPU or unified CPU GPU combination. and Humayun: what they do is they use that customized soul agent laws memory system Humayun: to have their own personality according to what I've designed, and they can Humayun: give me information based on things I've done.

For example, s one of Humayun: some of my agents, and you know, I think I might go into Humayun: Quite a bit. We don't have enough time to go talk about it, Humayun: but some of the books that I read, you know, and the ideas that Humayun: I have while I'm reading a book, I can feed that that agent Humayun: those ideas from the book.

And what it does is it combines all the Humayun: different ideas from the books I've done, and any decision I have to make, Humayun: it puts it through that and kind of gives an output based on Humayun: everything that I've discussed with it. So there's a lot of ways to kind Humayun: of customize these, and I think people need to learn how to do that.

Humayun: People you know, because either AI will make us all sound the Humayun: same, or AI can eventually make us even go deeper within our personalities Humayun: and our and our personal skills and and extract more out of each Humayun: person and and kind of highlight their unique individuality. So I think it's Humayun: up to us and how to use it. That's why more and more Humayun: people need to learn how to use these.

a lot of the college Humayun: students that are going through these to these higher level education system, they'll Humayun: they're go diving deep into this, they will know a lot of these. Humayun: And how to create personal agents.

And then what's going to make you Humayun: stand out is how good your personal agent is at doing a certain amount Humayun: and what you teamed up with your personal agent, how well they can accomplish Humayun: a task compared to someone who's just using basic ChatGPT or Claude. So I Humayun: think that's where the world is going, but we'll see. I may be wrong. Adil: I like it though, man. Humayun: Yeah. Alright. Well that's it. Well, I have Humayun: to figure out how to fix my car.

Adil: Alright guys, we'll tune in next time. Adil: Yeah, let's let's call AAA, baby. Alright, take Humayun: Yeah. All right. Adil: care, y'all. See you next time. Two docs, one mic.