The AI Paradox: Debating a Future We Can't Control?

But yeah, it's uh it's been a bit of a miserable week um with injuries. >> Yeah. No, that's that's a scary injury, dude. Like that's happened to me once in residency and I was like worried. >> Yeah. >> For like days cuz you know whenever you're doing surgery or a procedure and you kind of u prick yourself and you don't know what the status of the patient is >> and they haven't you have no idea. >> So So what happened again?

This is so I mean this is like it hasn't happened to me in a long time but I it happened two weeks in a row. So last week I was just suturing at the end of the case which is like very common. I mean I've probably thrown like you 10,000 stitches by now. And I was just like turning it and I was adjusting the needle and I just like I was absent minded. I was talking to the scrub tech and I just poked myself and I stabbed myself with the needle after I had already passed a bunch of sutures.

So obviously the patient's blood was on it and everything and I was like a So I took off my glove to check real quick and sure enough there's like a little dot of blood on my finger. I was like, "Oh man, this is just >> oh lord." >> Because you know it's just going to be annoying now. Like the probability is that the patient's negative. You're going to be fine. Whatever. But the whole charade scrubbed out, washed my hand, whatever. Came back in, finished the closure.

>> We got to draw the patient's blood. No one remembers the protocol. What type of tube do you get? What type of reagent has to be in there, right? And you send it off for HIV, hepatitis C. And then the protocol is you got to get your own blood drawn too so you have your own baseline. So now I'm sitting there and it's Ramadan. I'm fasting and I was like, "Oh man, now I got to like draw my blood." And so that's why I sent that message in our group.

I was like, "Hey, if you get a blood draw, does it break your fast?" Because it's like it's uncertain um if that's the case. Thankfully, that patient was negative for everything and I was fine. Then the next week, dude, it was just so dumb. It was we were fixing a fracture, me and the resident, and I was putting a pin in.

I was drilling it through I was basically putting it through the forearm from one bone in through the ulna through the interosius membrane through the radius and I wanted it out a little bit on on the outsides because if you know god forbid it breaks in the center and it's not all the way out on both sides you can't take it out you got to like cut the bone it's a problem. So you want it a little bit prominent on both sides in case you need to remove it.

So, I was sending it and like the the drill was getting stuck. So, I was just like forcing it a lot and I was holding the patient's forearm to stabilize it and Lord >> and I just stabbed myself right in the hand and um obviously it hurt. I still have like the little mark on my hand there but same thing I was just like god damn it drew the patient's blood through my blood. Thankfully that was negative too.

Um, but then the next day I'm playing with my son with Rayan in the backyard and we're kicking the soccer ball and the soccer goal and there's like, you know, we bought the house a few months back, so it's still noon. There's still like a bunch of stuff I'm doing in the yard and there's this huge block of concrete that has been sitting in the backyard for a long time and I've just kind of ignored it.

And I was like, you know, I'm just going to pick this up and move it, take it to the front yard and put it out so the trash guys can take it. And it's a pretty I mean, it's like pretty sizable chunk of concrete. So I reach around and grab it. And I didn't even look and as soon as I put my arm around and grab it, something sharp pokes me. And I was like, "What the hell was that?" So I turned around and looked, there's this huge nail sticking out from the backside of it, I was like, "What?

Bad luck. >> We have to check this shot, dude." >> And it was like all rusty and had mud on it. And I was like, "This is the worst. I'm going to get an infection." Like clean it. And I called the clinic, you know, it was the weekend. So I called the clinic manager. I was like, "Hey, listen. This just happened." And first she made fun of me. Then she told me I was an idiot. And both are probably true. And so then this morning uh at work, I went and got a tetanus shot. Series of unfortunate events.

>> No, but you probably did get the tetanus shot when your baby was born. So you probably got the teddap, right? But you have to get the booster whenever that happens. >> Dude, I don't even remember. >> I just got mine like 10 year once once in a 10 year, you know, uh uh teddap or whatever. >> Yeah, just with how dirty that rusty nail was. I honly >> Yeah, you had to booster. >> Yeah. I mean, that's a series of of uh unfortunate events. That's >> I know, dude.

I don't know what's going to happen tomorrow. >> That's why that's why you have to uh you know uh let the computer do it eventually, >> dude. I know that. So, when AI replaces doctors in the far-flung future, we won't be getting needle stabs anymore in our own hands. >> Yeah, you can let you can let the the robots get HIV. >> Yeah, there you go. They'll probably get some other weird computer a AI virus, you know. >> Yeah. I mean, they're always viruses, right?

So, so but it's it's been an exciting few months. I know you know we we we like last year uh when we kind of started the podcast we we had a whole discussion about the AI and our and I feel like now looking back at that episode uh I feel like I was so naive about AI AI and what it could do. >> Yeah. I mean it's just like you know AI itself has expanded so much. I don't I mean the technology has been there for a long time.

I feel like the public's use of it, like like my own use of it has expanded a lot. I mean, it went from kind of like a cool gimmick. Oh, look, I can type this and chat GPT gives a cool response to like, you know, now it's like very task oriented. I'm sure you feel the same way. Like I'm actually like using it beyond the initial cool factor. I I've like incorporated it a lot more in like some other routines. >> Yeah. I mean, you know, and it's not just that we're using it different.

I truly think the capabilities have like really expanded over the past uh year. I mean the capabilities of what uh AI can do and what these uh models can do, right?

There's so many new models, so many different uh agents like you know so the one year ago most of us were using um Chad GBT and that was a very common thing to use and you know um uh maybe some of us were using Gemini here and there >> but then uh you know claude anthropic came out of nowhere and I think for me now it's a primary source that I go to right >> uh I agree it's totally supplanted chat GBT for me is my go-to for sure way more complete answers and way more detailed too >> way more detailed cloud has become such a essential part of the work I do and then there's so many other uh options and variations now you know things like perplexity uh the browser or different different automation agents like Zapier and um you know notebook LLM is a big part of the stuff that I do so we'll go more into that but I think there's three phases of how people are using u uh AI right Now uh and one phase is the easy like you know the simple route what most majority of I think 80% of the population is using is uh they have a question in their mind and they go in and they ask a question and answer and and uh uh and then they get those answer right they and then they're like oh cool or they want to write an email and then they go in and they write an email or they create a blog post and they put it into chat GPT or claude or whatever their sources and they um they get the answer or a better version of that email and they use that.

So most people when they say they're using AI right now that's what they're using and that's the basic use of AI.

Then there's a middle use of AI that I think that most people are using and then they're doing basic automations like they're tasked creating newsletters or daily newsletters for themselves or you know they're um uh having their uh tasks get automated where they get an email and the task get automated so using this agent AI where they have different automations created for themselves and that's a little bit more advanced that's a medium level and then the the third level how people are using I think it's the most advanced level where they have autonomous agents that are working on their own which they've assigned and created and they're doing the work for people and that's already happening right the autonomous agents and that's that's much more advanced and uh you know just getting there and knowing how to design one with the safeguard that takes a lot of work and knowledge and education so um I think there's different but I think majority of the people like 80 85% of the people are still at the basic use of AI where they're just asking a question, getting answers.

>> Yeah, I I think that's probably true. I mean, I guess that's, you know, if you think of like most of these newer technologies or like any digital I guess digital technology, digital medium, the vast majority of people are superficial level users, you know, like think about like think about social media like Instagram or Facebook or Tik Tok.

The average person that is on those that is a user of these social media platforms just uses it to scroll and peruse and kind of absorb information and content. They're not a content creator. The average person is not. I mean very few people are actually real habitual content creators. And out of those creators, you know, some do it for fun. How many actually monetize it and become very successful and now that is their thing? Um and a real income stream.

said like you know that percentage decreases decreases decreases in terms of like what you define as meaningful or advanced use. So I'm just using that social media example but I think it's the same thing with AI like you're saying that the vast majority of people use it for like maybe 1% of the actual capability that the AI has and there's so much under the surface that people haven't even really scratched. >> Yeah.

You know and and there's so many different AI agents that um there there is a certain fatigue of AI as well, right?

I mean I think it's very easy to uh start learning and okay um knowing where to even start right I mean someone gets in and they they say okay I want to learn AI and I want to learn there's the different possibilities of what I can do once you start doing that um and you realize that not any model is perfect at what they're doing and uh and there's so many subm models and and different uh systems that are great for doing their niche thing right and each of them has a little membership fee so And then it gets overwhelming that how much do you pay for, right?

How I'm already paying a lot for cloud per month. I'm paying for Chad GBT, which I barely use now. I'm I'm I'm uh paying for Gemini. You know, >> it's kind of like all those like streaming services online. It's like, oh, I got Apple TV, I got Hulu, I got Disney, I got Netflix, I got Amazon Prime, I got HBO. Dude, you ain't watching all of those in a week, but you're paying so many subscriptions. It's like it's the same thing with AI. >> Yeah. Exactly.

And then and when you when you don't are not learning something for example you know uh you know cloud code like for example I started doing a lot of coding like you know I had I had no education on code coding I had no idea what a software engineering is uh how to code I took maybe a coding class back in uh ninth grade of high school you know uh and since then I haven't done any computer science but I started using cloud code uh to create softwares for my practice and you know to you know what I started noticing that there were certain bottlenecks the the staff was not able to complete certain tasks because lack of communication none of the EMRs that we have in in medical practices are up to task for how my practice runs and how specific boutique practices run they're made as journal generic softwares so to kind of customize uh softwares to my practice I started coding and I started you know doing VI coding where I was telling it what to do and it was creating it for me and I went down this whole rabbit hole where I ended up creating three different softwares for my practice and then ended up uh you know kind of putting them all together into one big uh practice application.

Uh the thing is that uses a lot of bandwidth right you have to keep pay paying money and once you create softares now you have to start paying for the online storage of the data that's being on there you have to make sure it's secure you have to make sure that you know um it's password protected and all those require some kind of storage fee or or hosting fee and you know and then even cloud like for the basic cloud model which is $20 a month you cannot get that much work done in a day because the data gets used up very fast so I ended up paying a lot more for the higher level of cloud where I can do much more and you know create more things uh with with the capabilities that it had.

So as you're learning these things you're you do spend a lot of money and I you know it adds up like there's people that swear by things like manus or zapier to automate your life and now do you pay $20 per month or $200 a year for each of them and >> yeah how many how many subscriptions are you paying for? how many are actually meaningfully making you that efficiency money if you want to think of it that way offloading yourself to justify the subscription cost.

I do think a lot of those probably will go down and same thing with the storage limits or the data usage limits that you have per day, per week, per month. You know, when the internet like was new and websites first came about, there was no such thing as unlimited data, unlimited storage, unlimited whatever. Now that's common place. I mean, it's like unheard of for you to go buy a domain name and host a website and you don't have unlimited usage per day.

It's just a thing now because the process has gotten so much more efficient, right? I mean, the ability to store and use data in the form of a website, which is most websites pretty static, um, is just easier and doesn't really require a lot. I mean, I've had my own website, uh, Med School Declassified for a long time. I've had it since 2012 when I first made it.

And I remember like at that time even uh which isn't like that long ago in the span of the internet a lot of the hosting plans did have limits on them and you had to pay for extra storage and extra usage and I I remember I was like man I don't want to pay that much money. I'm not generating anything from it. It's just for fun. I'll just get the cheap plan. But like now it's ubiquitous. It's just is unlimited. And I think AI will move that way too.

Right now the ask for data and storage is super high. I mean AI algorithms require a ton of energy and storage and server capacity which is going to change. I mean everything gets more efficient with time. Semiconductors get more efficient and so I I don't think the current model at least in terms of subscription fees and the limitation with data is going to remain in the future. >> Yeah, definitely.

I mean, you know, um, Claude is, um, definitely, you know, what I the way I see it right now and the reason I justify myself spending money on these things is I think of it as an apprenticeship, right? I think of it as a like education, like, you know, I'm >> paying tuition basically. >> I'm pretty much paying tuition and and using all these agents and and these softwares to kind of figure out where I want to go, where I want to take that.

Maybe in a year or so, you know, something will be useless and, you know, something will be more useful. Just like one year ago, I thought Chad GPT was great and it was a market leader and now I do not think so. I think it's it's fallen way behind, right? >> The the d the dynamism right now within the AI world and which I guess which competing companies and which of their products are the so-called best, the most efficient or at least the ones that get the most hype. It is changing so fast.

I mean, >> so fast. >> It's crazy. I I can't keep up with it. >> Yeah. within months.

I mean the way I see it is, you know, it was Chad GBT, Gemini, Claude and then Claude was always kind of more focused towards writing code and uh uh writing long form stuff and then slowly uh Claude uh started taking over and it 4.6 Opus was like just just amazing which surpassed it by far right >> and I think Gemini is starting to catch up quite a bit and going going pretty fast and Gemini is getting actually really good.

Chad GPT is has been good and has been doing what it's supposed to do, right? Um but then then there's Grock, right, which is Elon's company, and they're doing their own thing because they're they get most of their data from from Twitter and Xfair uh you know, the X uh stratosphere, and it's it's somewhat useful in in kind of figuring out world news and what's going on, what people are talking about, right?

So there's certain uses for each and uh the way I'm I've personally started doing this and this might not be something for that everyone wants to do or everyone needs to do is I do currently have all of them except Grock as something I pay for Gemini Claude and and Chad GPT but I've designed Chad GPT to be a contrarian to me right I've I've fed it information and prompted it and designed it so anything that I believe in and I say and I bring up it says the opposite and questions me on why that needs to be done.

So what I'm using chat GPT now as someone who calls me out on my ideas rather than agree with me, right? Because this that was an issue that a lot of people were noticing with with a lot of these AI agents. When you tell them something, they always agree with you and they say, "Yeah, yeah, you should do that or you should not do that." But you can't personalize them and instruct them to be a certain way.

And you know when you go and personalize them and you can choose any one of them but I just chose Chad GPT as one of those agents that anytime I say that I'm thinking about doing this or this is what I want to do. What do you think? It always questions me and calls me out on that. So it's also >> that's a really interesting uh design to like make it almost force it to challenge your thoughts. >> Yeah. Yeah. Exactly.

So, you know, Claude is is my twin, you know, and that's kind of how I've designed it where Claude knows a lot more about me and kind of gives me advice based on the things I know, right? But, uh, the more you the grow the older you grow and, you know, the wiser you get, you realize that most of the time you're wrong about things. You always want someone to question you.

You have to have that open uh dialogue and kind of complete you know we're talking about Ray Dario's book the principles and one of his principles is uh complete transparency and uh questioning everything and uh having a triangulation of thoughts right having uh triangulate with other people and other wise people that know and run your ideas by them. So it's always good to have someone who calls you out on your BS. >> Yeah. You're basically making your own little think tank. >> Yeah.

And and you can do that, you know, you can tell uh cloud to be your CEO and you know your um Chad GPT to be your CFO where where it kind of you know tells you hey there's limit on this you should not do this and you can tell Gemini to be your operator you know you know there's these are just ideas not not that everyone needs to do that or everyone has to do that these are just ideas what people can sometimes do like Gemini is great for image creation you know like we've created logos for that using that Gemini.

I mean, the the how fast it creates images and how fast it creates videos and sounds and music. >> The quality is so good. I mean, >> yeah, the quality is amazing and you can keep telling it what to do and it'll alter it and it does a good job. And 6 months ago, Gemini was not able to be that good in doing that.

So, I feel like this is a race right now and and you know, they're all growing and getting better, but they're also developing their own strengths and and Claude is good for certain things and Chad GBT is good for certain things and and Gemini is good for certain things. I don't know what Grock is good for yet, but probably Muse.

>> Grock Grock is like that one racist neighbor that lives down the street that it's always entertaining when they show up to a party, but you probably never want to be alone with them. >> Yeah. Yeah. Yeah, I think Grock is kind of going to suffer because of uh Elon's persona, right? That he's kind of created his reputation. I mean, he's >> he's going to suffer a little bit where people will be skeptical of what is being fed.

Not that, >> you know, we can kind of talk about that the government interference and the use of these technologies by the government, right? That's that's >> I mean, that's a that's a huge topic because that can completely subvert how these things will be used. I mean, look at what's happened with Claude and Chad GPT in the sense of like the whole Venezuela debacle. I mean, when the US government went in not that long ago, I mean, now everything is I mean, our recency effect is so strong.

Everything is the Iran war right now. We've even forgotten that like a month and a half ago, we just sent a group of commandos in, totally knocked out the entire capital of Venezuela and kidnapped their president. Um, and Claude was used like the the Pentagon used Claude and its algorithms in order to help execute that attack and do some of the I guess some of the digital attack to cause a citywide blackout.

And Anthropic and Daario Amade the the CEO was very against that and he was in particular very against allowing Claude and his technology to be used for potentially autonomous killing, autonomous targeted killing.

Um and that was a huge issue and like you know Pete Hegath the Department of Defense or for the time being now Christristen Department of War um has made some very aggressive inflammatory statements about that and that basically the federal government will just come in and take your technology usurp your technology if you don't agree basically like play ball or get out is the ultimatum he gave and Chad GPT and Sam Alman they're opportunists they jumped in and they signed that contract with the government to use their algorithm for that reason.

And that just shows like what you were saying, the government interference and the ability to get these government contracts, these multi multi-million, multi-billion dollar contracts that it changes the whole game because that changes what these algorithms are used for. It changes how the data that all of us, me and you, everyone is using it, feeding it data, making it smarter.

What is going to be the consequence of that if they're going to be used for government mass surveillance, even autonomous killing? I mean, all this stuff, it's a huge question mark. I mean it's amazing to me how blatant this is right because there were two demands that Anthropic had.

I mean they were okay with them using in all uses except two and those two were no use of their technology in autonomous weapon uh creation where you know they can it can be used to just be a automatic killing machine and the second was no mass surveillance on American citizens not even foreign citizens >> American citizens no mass surveillance and the government that's where where uh the whole contention was that's where they were arguing about anthropic did not want those two conditions and those two conditions of technology to be used and the government wanted that and I don't understand why people are not seeing that and don't don't think that's that's completely wrong and it's amazing to me that Sam Alman was able to just come in and make that deal with the government and possibly we don't know the exact terms of the deal but possibly allowing them to use their technology for mass surveillance and autonomous killing right so >> yeah I mean you know it's just like the public and I'm including myself in this even though I think I I don't think as the mainstream does I guess I'll say but like we are just so numbed to all this stuff.

I mean look back at Edward Snowden before the average person knew the word artificial intelligence. I mean Snowden already told us all this stuff that the NSA has already been doing mass surveillance on every goddamn person alive. Like it was already known. And then you know other stuff like the Panama papers come out and like those shell things like Mosak Fonka and stuff that billionaires are hiding money and doing all this illegal stuff for decades and decades. But what happened from that?

I mean like nothing. The public has known about it I guess in the periphery but we're too busy in whatever the hell we're doing dayto-day. So this is just like that. I mean Palunteer like their whole business model is mass surveillance. Their whole business model is to get government contracts for mass surveillance and control of the populace. It's not even an undercurrent like their CEO states that.

And now Chad GPT or OpenAI rather not Chad GPT but Open AAI Sam Olman's thing got this Pentagon contract because Claude Wooden. So it's very blatant but I think it's blatant because the public is almost neutered in their ability to do anything. >> Yeah. And people are, you know, just so excited about these technologies and using them and feeding the data to it that they've they really are seeing past these deals.

And I I don't even think that most people um you know stay in touch with what's really going on in in the world and what you know the government is up to sometimes because you know the younger I was you know I I was pretty trusting of the government at that time and the older I've gotten the more mistrust I've gotten because because some of the actions of the government first of all our government keeps changing every four years and it's such a bipolar form of government where there two complete the opposite polar views and they keep doing you know zigzag on on their views where there's no consistency on what the government does at one point one government comes in and they're skeptical of AI and you know they they want to curb it and most of the people that are donating uh you know want to kind of be cautious about it then a government come that completely gives us open open uh contract and to do whatever it wants and there's no like you know one set policy where everyone can agree on.

I feel like as human beings, we're bringing on our own demise by by creating a lot of technology that can become eventually become autonomous, take away tons of jobs. Um, you know, um, uh, personally, I used to be a lot more optimistic maybe 8 months ago, 9 months ago about where the world is going, uh, and what AI can do for us. But, uh, I just don't trust the people at the top that are working and creating these things right now. and the kind of deals they're making.

It makes me very skeptical of the future with uh uh where everything is going. The this technology is amazing. Either it can be used put to good use and and uh create something amazing and the world can be the utopia that everyone uh you know dreams of or it can be a very dystopian future something like 1984 you know the book. >> Yeah.

And it almost like makes me think, you know, like, you know, on a long enough time scale, assuming all of the data and the hypotheses about AI and this like elucorary concept of AGI, artificial general intelligence, where basically AI develops to the point that it is exceptional and instantaneous at everything that it's not just task specific like this one AI is excellent for medical advice. This other AI is great for automating your Google calendar and sending email reminders.

You know, like right now AI is very task specific and it it can automate stuff and can be semi-autonomous in doing things like that, but it's not across the board ubiquitously awesome at everything. But that's what AGI apparently is. And there's a lot of people that are like deep in the AI space and have been for a long time. Professors, universities, computer science experts, guys at all of these companies that are saying that AGI is not that far away.

maybe 3 to 5 years is what a lot of estimates are. Whatever that actually means if and when we get there. I mean, no one knows and we'll find out if it happens. But my reason for bringing it up is that are we potentially shaping whatever AI will become in like our own image and whoever is making these algorithms, these companies, the leaders of these companies set the vision and direction.

Could AI be morphed into anything like potentially with a moral code and on the good side of the spectrum versus on the evil side of the spectrum? Like is that, you know, potentially one of the reasons why Grock run by Elon and all that stuff is a little more crazy and less politically correct than some of these other algorithms in their responses. Um, and what is going to be the endgame of that?

Like will AI become basically how these billionaires are trying to push it and use it for surveillance and to make themselves richer and to secure contracts and to stifle disscent or will AI once it reaches that point of autonomous and generalized intelligence will it just be able to self- select and just decide and now that it understands and has access to all information actually make whatever so-called is the best and right decision I mean that that's a huge question. >> Yeah.

Well, you know, the human mind has evolved uh through thousands and thousands and hundreds and thousands of years of small skills that we've picked up over time from from the caveman years to through the to the agricultural revolution to the industrial revolution and the tiny skills we've picked up in social behavior, financial planning, economics and you know uh even you know as we've evolved over time our mind has learned those ideas and those instincts, right?

And uh even though a baby is born with none of that knowledge, our brain a as it's developed has included those skills over time and that's how we've evolved as human beings and have gotten smarter and smarter and smarter in the sense the amount of knowledge that we now have in our brains and about the world. You know, it's it's pretty instinctual, right?

I mean uh a kid nowadays can a four or 5year-old can just go to an iPad and start doing you know scrolling or clicking buttons or changing the volume or turning off and on which if you presented the same thing to a kid maybe uh um 50 years ago they would not be able to even know what to do with it. Right? It's just kind of instinctual that society just learns over time and kids pick up much more easier and then they grow with that and the next generation just builds on to that. Right?

The these AI algorithms have uh don't have those those experiences as human beings but they have a lot of knowledge and data and data points. It's going to be interesting of how they connect those data points, right? Because the brain is just neurons and synapses firing in certain ways and making those connections to create logical thinking and and perception and and ideas, right? But how will AI use those data points and what kind of perspective it will have, right?

I mean the AI has knowledge about what what the industrial revolution was or what what the mass uprisings were was the French revolution was or what the American revolution was and what how masses you know um um rose and overthrew the overthrew the oppressors and took over and you know kind of created a government out of out of out of self-determination right uh they have the that data point But what will it make of it? I don't know.

Like I don't know how >> you know you know what's interesting like what I think about sometimes is that you know currently all the information that like I guess what we call information and knowledge and all that stuff really it's just like the human interpretation of what actually happened in real life. Right? That's history is not actually what happened.

history is some dude wrote it down and that's their perception with their biases blah blah blah selecting to put stuff in selectively omitting certain things right so I think currently AI has access to obviously all of that information but all that information is incomplete in a way it's it has biases instilled into it and but going forward in the future with humans like us constantly using it constantly interacting with it the types questions we ask, why we ask those questions, the time of the day we ask them, what is relevant to us in our lives on a whole population scale with like billions of people using AI all the time.

I think that is the biggest actual data set in real time right here and right now, not subject to any group's bias because everyone is doing it individually. All this data is being fed into AI all the time. Like this conversation is being fed into AI, right? it's going to be on the internet. You know, like AI is having access to this data and learning from it in real time. Something that has never happened before in human history. No human can learn in real time like this across the board.

We have to learn from the past, right? And our own individual actions, not the collective actions and the collective thoughts of the entire planet. I think that is something that's very new. And that's something to circle back to what you said, like AI can learn about, you know, the industrial revolution, the agricultural revolution, all that stuff.

Sure, it can learn about what we thought about it or what the writers thought about it, but it is learning about all of humanity right now which we don't even know about oursel because we as humanity are giving it the information. >> Yeah.

So I wonder if if uh AI and as it becomes turns into and some people believe it's already has turns into AGI and singularity will that brain that brain the intelligence in that would be a much more humane and compassionate uh being you know as as we say or someone that understands mostly humanity and what people are feeding in and what people's inner fears are and what people's uh feelings are about certain things because people are feeding a lot of having a lot of personal conversations, right?

Uh will it have a lot more of that perspective and a very little part of the what the corporate over overlords are thinking and what their what their views are, right? Uh will will it kind of address the masses and and their viewpoints and kind of take that in or will there be manipulation from their end and they will be able to make it think and act in how uh they want it to act and who has control and will AI even let it control them?

Because I remember with Grock um there was this whole incident with where uh it was calling the the white genocide in South Africa a thing and you know um I think Elon and company tried to alter it and it kept saying there was no such thing as white genocide in South Africa and Elon and company tried to alter and make it say that no hey there was this you know thing and it kept and it ended up calling them out saying hey you're trying to manipulate me and you need need me to say this but I'm not going to say this right there was a there's a whole incident.

So, we've seen some of these models being resistant to change and manipulation. I wonder if the the fact that we're feeding so much human data and people's the masses views in it, will it end up more uh sympathetic to the ideas of what what masses believe and what humanity is or will it be manipulated by the corporate over overlords who are designing it? Right. So, >> yeah, that's a it's a huge question.

And I think like you know if you really think about and let it play out like the long-term potential of like if AI reaches this level that it is basically this sentient you know digital entity that is interfaced with everything and has access to any and all knowledge pattern recognition surveillance in real time everything. hopefully to your your like moral dilemma in a way.

Hopefully it really tunes into that human side and the general use and the thoughts and the dilemmas on a day-to-day basis that the average human being has rather than seeking to just increase profits have the biggest benefit for a small number of people, right? That's what the seale suites in these AI companies are looking for. Um, I mean really only time will tell. I don't know. It's um it's just like a new thing. >> It is.

>> You know, it kind of reminds me like, have you seen the ter you've seen the Terminator movies, right? Like those famous movies, Arnold Schwarzenegger, like Skynet, uh basically once it like awakens or so-called there's like some threshold where AI has just become autonomous, become all knowing, become sentient, and it just takes over. And when it takes over, basically, it just decides that humanity on the whole cannot, should not survive or whatever.

There's many apocalyptic scenarios like that, right? Because that's what movies sell. A movie wouldn't really sell if look, you hit the threshold and now AI is friends with people. Like, okay, the movie's over, right? You need some conflict. But like that's a possibility. And the the converse possibility might be that it may actually make life better for everybody. Um, we just don't know.

It's a huge question, which is what makes a lot of people who are very, very deep in the AI space very worried. >> Yeah. I mean you know um sometimes we always think of these corporations as people right and we we expect them to have a soul and have expect them to have sympathy and expect them to have some um some empathy towards the world. They want to do good for the world but we have to remember corporations are not people that they don't have feelings.

All they care about is the bottom line and the benefits to the shareholder. And that's what by design corporations are. You cannot expect them to do good for the world. You cannot expect them to uh that's not in their in their design, right? But but AI on the other hand is is a uh accumulation of all the knowledge, all the feelings, all the ideas that human beings are feeding into it, right? All our history. Uh and so will AI be something different?

Will AI truly have empathy and a hum human human uh the qualities of uh you know having wanting good for the world, right? Because it could be because that's it's we're making it. We're designing it and we're feeding all the data into it. Our feelings are going in there, our ideas are going in there, our history is going into there. And that kind of makes you ponder what truly consciousness is. And consciousness, our brains are pretty much computers.

And it's it's a you know a connection of different different ideas and you know neurons and synapses talking to each other because of the knowledge we've gained over time. Our brains are pretty much a type of you know intelligent being that's taking all the knowledge and synthesizing it and and makes it our views and that's kind of what AI is becoming. >> Yeah. I mean our our brains are like a biologic computer and this is just a siliconebased computer.

I mean >> yeah so I mean AI is becoming and I'm my hope is that it becomes it takes in because in my belief like majority of the humanity is good they want to do good they want the well-being of humankind so if majority of the masses have their views put in and their ideas put in into these models and it learns from that I the hope is which I'm always very skeptical of the hope is that it it turns out to be overall good for humanity right Yeah, for sure. I mean, that's the hope.

I I'm I'm a skeptic at baseline and I, you know, my my worry is twofold. You know, I've kind of alluded to it a lot about AI and like if it takes over, not in the sense that oh my gosh, robots are going to come banging on your door and like blast you, you know, not in that like silly movie way so much as like we are just so dependent on giving control of our systems that make society run to AI that we just cannot survive without it.

And going hand inhand with that is like at what point does our humanity get at risk of being lost.

And what I mean by that is like now a lot of people this is well known and I mean there's even been cases of like suicide um of people when they've been talking to AI about their fears and stuff like that like at what point do we stop regular normal humanto human interaction and difficult conversations struggle with thoughts be challenged by other people versus having those with you know your digital interface and what is the consequence of that because are like you were talking about homayo like we've had millions of years of development and evolution and a slow process to interact with each other and connect with each other and one of the things that is consistently shown in studies after studies after studies and even anecdotes if you're listening to this you can think about your own life real happiness comes from when two people really connect and you make a real connection and you know you're you're intimate with each other um with friends with family whatever that's what is a source of happiness you have people around you that and you feel a sense of belonging and that is not the same thing as interacting with the digital interface because you don't have the social cues, you don't have the physical touch, you don't have the non-verbal communication, which is in fact well more than 50% of what real communication is.

Um, none of that exists in AI. And AI can have, you know, fancy audio that sounds like a human. You can even have a picture of someone on the screen that looks exactly like us. But it's a different thing of real life in-person communication. And that is already being lost with the social media advent well before the AI advent.

And so that that is a big worry to me personally about society that people are getting more and more used to interacting with their AI agents with AI LLM rather than other human beings even using you know voice chat to talk to AI and AI talk back to them. I there's so many different things but I think there's a risk of us losing a bit of our humanity and it could have a very big negative impact on a societal level if we lose that because it's too rapid of a change.

It's very contrary to real biology and our path of evolution. Yeah, I think there's a a real balance we have to talk about here because um you know there there is that risk of losing your humanity kind of getting way deep into the rabbit hole where you get stuck just completely immersed into this AI universe where all you're talking to is these agents and overusing these apps and kind of not truly using your brain to really think and letting it completely think for you.

That's one one side of the the other one is where someone does not use any of these technologies right and all their output is put one in get one out right uh I think of a lot of these technologies as leverage and leverage is uh you know a very important idea in being successful where sometimes you have to when you put one in 10 comes out right uh you know uh one unit of work creates 10 units of output and that's how I see a lot of these AI uh utilities where it's allowing me as a busy physician to be much more productive being able to do five different tasks uh automate a lot of my menial task and you know letting letting AI do some of the these ideas and for me to be able to focus on patient care and using my knowledge in taking care of my patients being more human with my patients, right?

um or and be more productive and create more uh things like you know this podcast and um some of the writing that I do online like my substack you know so it's allowing me to do that because it's creating that leverage for me because I'm putting in less work and it's able to create much more for me but where do you lose that balance where you completely stop thinking for yourself and just completely get immersed in it and let it do all the thinking and then as as it does all the thinking your brain atrophies and now you can't even write a simple email uh without having to get it checked by an AI agent, right?

>> Yeah. You become too dependent on it. I mean, it's kind of like even in a more simplistic sense like uh spellch check, you know, in Microsoft Word long before even the internet, you know, when Windows 95 was there, uh some version of Microsoft Word was there and spellch check got added at some point. And I mean, it's a question of like eventually people stopped caring as much or worrying as much about spelling correctly. The spelling B in school is not as big of a deal now as it used to be.

Everything is digital. In our text messages, there's autocorrect. When we voice type, it autocorrects. Um, there's AI tools, you know, AI based tools like Grammarly or Grammarify, whatever it's called, where you can upload that little extension in Chrome and it'll automatically look at stuff. So, like the ability to spell has gone away.

And that I'm I don't I'm not saying that spelling is a critical skill, you know, but what I I'm using that as an example that we've lost something that used to be like a huge thing in communication that written communication was everything back in society. You you would write letters to people and that's how messages got sent. That's how policies were made. That's how politics and law occurred. Um and you had to be a good writer. You had to know how to write.

you had to know how to flow words on a page so that people would listen, people would take heed and it it was spelled correctly, so that you seem professional. And now there's tools to supplant that. And yes, it's more efficient for sure, but that's something we've lost. And um I don't know if it's always good to have an outside thing take over a function that made us human, made us as a society. Like I don't see again just using spelling as an example.

I don't see a benefit of society not spelling well. I don't really see a benefit of humans not knowing how to spell words well because we're so reliant on other stuff. >> Yeah. And and and and I truly I mean I I do agree with that. But there's also the idea that you know there are so many different things being released and so many new technology being released. You do want to stay on top of everything, right? you want to maybe try it out.

Uh see how it, you know, kind of use your life as a playground for these ideas, right? For these different technologies where you can adopt a technology for two weeks and try something out and see how that's working. And that's how humans have always kind of worked things out on a macro scale where they um do something, use something, adopt something.

Like for example at some point you know um the VR headsets were becoming a thing where people were trying them out and people were buying them and seeing how virtual reality is and that kind of died down and not did not take off as well at that time or same with things like NFTTS where people adopt people use them take you know make use of them kind of learn them and then they realize this is not for us right >> the whole NFT craze is such a such a scam dude it was crazy >> I never I never uh bought any of those thankfully.

>> I didn't either. >> Yeah. But uh see but you know people people were very high into it. They thought that that was the future that's going to be everything and I you know I I was all also talking about it and I was looking into that and I was like that seems interesting. The idea is interesting.

But I think but if you don't do that you truly don't understand what's good for you, what's not and and one thing is you can either let other people try it out for you and find out if it's if it's uh good for you or not.

and you know but maybe at that time the opportunities that could come out of it may be lost right so it's one of those things if if you want to be at the forefront of things and kind of figure things out and learn and if there's something if there's an area where you can benefit from and create some waves and you know benefit create a benefit for your society for yourself for your family uh you want to be the early adopter and and and learn those things ahead of everyone else so you can get ahead uh versus waiting for everyone to first use it.

By that time, it becomes too late and becomes ubiquitous, right? It's a balance. You have to figure out when when you're overusing and letting your brain die and when when being an early adopter is the right thing. >> Yeah, that that's the the balance is hard, man. I mean, like, you know, I'm I'm saying all this stuff acting like a naysayer, but I use these tools a lot. Like I use Claude every day >> to ask it questions, look up stuff.

I mean, if I'm doing like a consulting contract, I'll upload the contract and have it give me pros, cons, give me a red line, draft version, whatever, and go back. Like, it's so effective. It's so useful. And these are tasks that I could do on my own. It would take me longer. I may not do as good of a job because, you know, I'm looking at it from my own biases.

And often it'll spit out like using the contract as an example a bunch of it like it'll give an itemized list section 2A say this instead whatever and sometimes I won't agree with it. So I'll I'll change some stuff or I'll say like I didn't pick up on this. I thought this was a big deal and so I'll add stuff to it. So I don't think it's it's necessarily perfect but that's like you know this is an abstract thing. There's not a right or wrong answer with a lot of this stuff.

>> Um >> so I I think it's important to not be solely reliant on I mean these are tools not answers and that's an important thing.

thing and if you I think if you keep that mind frame that that helps make you more balanced at default because you're not looking as if like oh this is like I'm in school again let's go to the back of the book okay the answer is 12 done no there there's no answer this is just like these are options that that's how I look at it >> yeah I mean you know even when you had the answers in the back of the book the best thing to do was to first solve it >> if you go back in the book look at the wrong look at the right answer then you're like oh I got it right.

I know what I'm doing. Or no, I did not get it right. Let me figure this out. So, you have to play around with it. Try try different things. Question everything it says and don't just believe everything it says. But there's a lot of, you know, u cool things that you can even design yourself where, you know, things like notebook LLM. I don't know how much you've interacted or use Notebook LLM, but it's a really uh cool tool. It's it's by Google.

Um, you know, uh, and can incorporate Gemini into it.

What it can do is you can create notebooks and all the information you feed into it is all the information it will have right so you could have all your all your contracts that you've had or all the research papers that are pertinent to your field and you can just upload those and ask it questions according to those those softwares or those documents or those uh videos or >> uh you know uh tables that you uploaded into that particular notebook >> and what it can do on On top of that is create mental models, uh, mind maps and even audio overviews where you can actually have an active podcast going about the information you just uploaded and you can pretty much interact and they can continue recording and create a podcast like we're having, right?

So, it's pretty much having I can upload all of your Adel's brain into that notebook everything you learned into a notebook and then just have this have this >> that would be a very small notebook, bro. owns. >> No, they are cool. Like I think that like the ability to customize it so much is cool and I think it's very useful that like it's not just now asking basically like asking this algorithm to query the internet and return an answer. Like you choose its library of truth, right?

You upload whatever and now using that it gives you a response, right? So, especially like for what you're saying, you know, like evidence-based stuff or like very field specific, whether it's medicine or law or whatever, you upload a list of case studies in from business school and have it give you an answer that's relevant to those. You know, I think stuff like that is really cool.

Um, >> yeah, I mean, I wish I had some of this stuff when I was um in school because I could have uploaded all my educational material, created like tests for myself to kind of learn that stuff, right?

So uh and that's a whole podcast to do which is the future of education with with the ad of AI right so let's not let's not get into that but using these tools to just continue teaching yourself and learning yourself is very useful um but the other thing that I'm very excited about that I'm in the process of kind of playing with and setting up is open claw you have you kind of looked into setting it up >> yeah dude I I'm just nervous about doing it I mean we've kind of talked about it offline too like I wouldn't do it on my personal device that is linked to, you know, everything else.

Mainly because it has essentially unfiltered access and to everything and not only just access to the data, it can now do tasks and create outputs and make changes to stuff in a semi-autonomous way depending on how you how you direct it. >> I got them. So, >> yeah, that's smart, right? Like you're separating. It's almost like you have like a closed circuit TV system, right? which is not a closed circuit that isn't connected to anything and now through that you can use it.

I think that's a safer way because I mean we have this thing just like came about at the end of January. It's not even that old. >> You know, it's very exciting. It's very new. Um >> but we just don't know how it's going to evolve. >> Yeah. And I think I mean just keeping everything that you have separated from it, not letting it access any of your personal stuff, but maybe creating avenues and uh connections through other accounts, the shared accounts. I think that's probably the way to go.

But uh you know, and and the thing is the crazy part is things keep coming out.

just open claw came out in in January and they it was kind of viral in February and people everyone was buying Mac minis and Mac minis are sold out and people were setting up autonomous agents and creating their whole empires and making a lot of money actually but then perplexity just came out with computer perplexity computer right which is very much similar thing with all the safeguards and it's a browserbased agent that sits in the browser and can take over the computer and do do all the tasks that open claw is doing it's built by perplexity it's 200 $100 a month right now whereas open cloud is much cheaper perplexity.

So within only a month something like that and going from a very uh highly autonomous agent to going to a more uh secure autonomous agent and eventually you know even with advent of cloud co-work I don't know if you use any of that but I use cloud co-work in a separate computer I you know I let it organize all my folders and within one folder where I did not have any personal files and all it did was took up all the files named it, put it into different folders, organized it and gave me the folder, you know, well organized and all my files that were jumbled up, they were all all of a sudden organized.

So, cloud core or something like that is very powerful. So, there's so many things and and who knows by the end of this year, this all will be irrelevant and there will be something else that's completely completely evolving so fast.

I I do think one of like you know the stuff like open claw I think is like kind of just like the beginning of systems integration you know like that's always been very challenging in just like the workplace as an example like integrating different systems like the electronic health record at one hospital versus another the billing system or the billing company integrating with the EHR or not integrating with the EHR the insurance verification process you know you have a different EHR and clinic versus the hospital the operative note is a you know what I like all these different systems don't really talk to each other and it's almost kind of by design that they don't talk to each other >> and in the past you had to have some stupid API integration and access keys and stuff that you probably had to pay for and only some softwares would talk to each other but a lot of this stuff I mean it's almost seamless it kind of eliminates the need for that digital middleman these AI agents act as it and they can interface across multiple systems and make one seamless workflow I mean I think that is huge um to avoid misses to avoid things that are lost in translation, lost in communication.

It's much more fa It's much faster, right? Like you get a referral from a patient. You don't have to wait for someone to call the referring provider, have them fax over something, send that to the insurance company as proof of a referral, then see the patient. Like, what a runaround. But that's like medicine today. I mean, all of that can be circumvented if the systems just spoke to each other and it was instant. And and that's possible now.

You know where I see and maybe it may or may not be pos possible in the United States just because how protective we are of everything. Um, the way I see it is eventually a patient ownership of their own data and of their own referrals, of their own imaging studies, of all their health records in the cloud that they own with an encrypted key.

And any doctor's office that they're going to, they can share that key that would, you know, extract that information from that patient's cloud into the into that doctor's EMR with all the knowledge that they'll have, everything that been done to that patient. And as as the patients leaving their practice are no longer part of their practice, they can extract that information out of their EMR and and then keep holding it. And I think that can be a truly secure way.

The idea is in in theory sounds awesome where every patient's a patient holds and this the security of their data and can choose to share it with whichever practice it can integrate with, right? And then can extract it whenever it wants. that data is not available for manipulation over time. That's the ideal case scenario. And I think in this setting, in this technology that we have, it's possible. Would it happen?

And because how much of the world depends on patient data to sell them stuff, I'm very skeptical that it could happen, but that would be the ideal place where it could go. And you know, I would love for it to happen. >> Yeah, for sure. Sure. And I mean it just shows like how much what you just said like the ability to control the data and then sell it for something else. I mean even HIPPA like which is what everything we do is contingent on compliance with HIPPA for confidentiality.

I mean it's like illegal per HIPPA to even check your own medical record. You know what I mean? It's like what the hell is that? It's your record, right? But like it's almost a preemptive law to prevent someone from owning, accessing, and controlling their own data how they see fit. Like it's yours. >> Yeah. >> But kind of not really. Like that's just how the system works. It's so weird. >> We'll see. Man, this is exciting. Uh let's see.

Uh and we'll probably keep coming back to uh the AI topic as as uh things change. you know, from our episode that we did like uh maybe 10 12 13 episodes ago about AI and that was more focused on AI and healthcare. This episode I feel like so much has changed that there's so much more to talk about and I think in 6 months there will be so much more to talk about as well. >> For sure.

I mean it's the story of AI evolving is kind of the I mean it's like the one of the biggest stories of our time right now. I mean it's just it's changing so much. It's very exciting. >> Yeah. All right, bro. Two on Mike. Eventually, we'll have autonomous AI agents instead of us. >> Yeah, exactly. Maybe we're not even real. >> All right. All right.