What we talked about
Shariq A. Salar, Founder & CEO of Atrean, joins Federico Ramallo for a philosophical conversation about AI, decision intelligence, and how we build. Atrean’s AI-native platform turns natural-language questions into instant, governed insights:bringing “lean intel & leaner ops” to investment teams and beyond. Together they explore non-determinism and creativity, vibe coding and rapid prototyping, why human abstraction still matters, and the shifting roles of juniors vs. seniors. They also tackle ethics (ownership of training data), social questions (loneliness and AI companions), and transhumanist ideas about identity and continuity:framed by the sci-fi stories that inspire today’s builders. A reflective, forward-looking dialogue on why we must build better intelligence:responsibly.
Show notes
Shariq Salar named his company after a Star Trek species with deep listening abilities, and revealed that was a coincidence only after the name was chosen. That accidental alignment captures something real about how he thinks: the most interesting insights in this conversation arrived sideways, through analogies and questions rather than declarations.
What we covered
- Shariq challenges the assumption that AI homogenizes creativity. His counter-argument: if hallucination is how AI generates novel outputs, the same way a person in the 1950s imagining flying cars was technically hallucinating, then maybe the most interesting use of AI is to build models whose explicit job is to hallucinate reasonably, to help us predict futures we can’t yet see.
- The democratization of software building is already happening at a pace most people underestimate. Shariq notes that fresh-out-of-college engineers who have never known a world without AI are outbuilding seasoned veterans at specific prototyping tasks, because they don’t fight the tool’s non-determinism the way experienced developers do.
- He sees AI’s most critical near-term application as defense against biological threats. His argument: if the incidence rate of pandemic-level events is accelerating from once per millennium to potentially once per decade, humanity will need a form of intelligence that can respond proportionally faster than human coordination allows.
- Shariq draws a direct line from early smartphone adoption to the current AI transition. Street food vendors in the developing world now use UPI payments and Google Maps, technology that would have seemed exclusive to IBM employees a generation ago. He predicts AI will follow the same trajectory toward universal integration.
- He pushes back on the idea that offloading memory to technology frees up cognitive capacity for higher-order thinking, but finds the question genuinely open. The Sherlock Holmes model (deliberately discarding irrelevant information to preserve space for what matters) might be a real cognitive strategy, not just fiction.
- On the question of social isolation and AI companions, Shariq expresses concern about the direction Japan is already demonstrating: if an AI partner can be molded to be perfectly accommodating while human relationships require negotiating imperfection, the incentive structure points somewhere troubling.
About Shariq
Shariq A. Salar is the founder and CEO of Atrean, an AI-native data platform that allows teams to ask questions in natural language and receive immediate, governed insights without traditional BI infrastructure. He has spent his career building and scaling companies, and he operates out of San Francisco.
- LinkedIn: https://www.linkedin.com/in/shariqsalar
- Website: https://atrean.com
Episode 34 of the PreVetted Podcast.
Full transcript
Federico Ramallo (00:03) Welcome back to the Privated Podcast, where we spotlight extraordinary people and remarkable talent reshaping our world. Today, I’m joined by Sharik Salar, founder and CEO of Atrium, an AI native data platform built to deliver clear, immediate decision intelligence. Sharik has spent his career building and scaling companies,
and he brings an operator’s discipline to a problem many leaders face, turning scattered complex data into trustworthy answers you can act on. With Atrium, teams ask questions in natural language and receive instant governed insight without the cost and complexity of traditional BI stacks or heavy analytics workflows. So thank you, Sherik, for joining us. How are you doing today?
Shariq Salar (00:59) Thanks Federico, glad to be here and thank you for that kind introduction. ⁓ feels great to be here. I’m doing well. How are you?
Federico Ramallo (01:08) I am doing great. I just came back from San Francisco, I’m getting used to the ⁓ rhythm of working from home again. So it’s a whole different dynamic, right?
Shariq Salar (01:19) Hahaha
It is, it is. And it’s pretty fast-paced right now with everything that’s happening in AI, don’t you think?
Federico Ramallo (01:29) Yes, yes, I’ve been to many meetups and meetings there where they talk about how AI is pretty much changing a lot of the new startups, right? So people were talking about that there’s so many green fields like in 2000s, right? Where there’s so many new opportunities of things to do now with the tools that AI is providing, right?
Shariq Salar (01:48) Mm-hmm.
Yeah, it’s a whole new world out there and I’m glad that I guess we’re in the right place at the right time. ⁓ In fact, in San Francisco lately I’ve noticed, of course, the place always has its energy and everyone’s always about building and being tech forward and being techno-optimistic, if I may say so. But right now, I think it’s a…
It’s a once in a lifetime thing to witness. I don’t think you’d be going to any meetup which has to do with, you know, could be healthcare, could be policy, could be, you know, governance or, you know, defense. And all they seem to be asking is AI. How do I get more AI? What’s happening in AI? Can AI do this? Can AI do that? ⁓ And so, as people who are building in this space, I think we’re very lucky.
that we’re in the spotlight, but it’s also a lot of pressure. Every time you walk into a meeting and you find two or three other people who are even thinking about the same things that you’re building or in similar spaces, it’s validating, but it’s also, it’s very interesting. Yeah, it’s a good time.
Federico Ramallo (03:22) Yes, yes, it’s very interesting. I think people are talking about AI as if it’s a mature thing. Everybody feels they project that they know everything about AI. ⁓ They talk about prompt engineering, right? I don’t know where I it about this. Instead of prompt engineering, we should call it ⁓ prompt alchemy because there’s so many things that we don’t know, right?
Shariq Salar (03:36) Yeah, yeah.
You
Federico Ramallo (03:52) Yeah, yeah, yeah.
Shariq Salar (03:53) That’s a good one, prompt alchemy.
Maybe he’ll get debunked as a pseudoscience someday.
Federico Ramallo (03:59) Yeah, yes.
I’m doing this ⁓ thing with the vector database of generating these ⁓ dimensions and doing this rack. I’m using, ⁓ I don’t remember the name, it’s a Ruby library, ⁓ graph something.
Shariq Salar (04:10) Right. What do you use, spine cone?
GraphQL.
Federico Ramallo (04:31) No, it’s not graph. Maybe I’m wrong. ⁓ But anyway, ⁓ we build… So I’m using PostgreSQL for a vector database because I’m already using PostgreSQL, so I don’t want to complicate the ⁓ infrastructure. it gives us like 1,500 dimensions, right? But nobody knows what these dimensions are, right? It’s like…
Shariq Salar (04:33) Well, anyway,
Mm-hmm.
Sure.
Ha ha ha.
Federico Ramallo (05:01) I don’t know,
it kind of works, it’s magic. So anyway, that’s where I am. And I keep asking everybody, what do these dimensions mean? And they say, people say, it depends on which LLM agent you choose or LLM engine you use, but at the end of the day, nobody knows. Yes.
Shariq Salar (05:25) Don’t
you think it’s a really strange way to be when people who have spent 10 years or 20 years studying coding and languages and data science and data structures and SQL, they’re being made completely redundant right now because now…
you don’t really need to know most of that to actually extract value out of most of these things. ⁓ So with most generative AI platforms or large language models.
or even these new AI-based tools that everybody seems to be using. It’s not the veterans that are getting the most value out of it. It’s actually people who don’t know how anything works, who are just able to instantly go and get results quickly. So you’ve got teams of engineers who are just out of college, who are building like, you know, really deep applications in different spaces that they don’t really… ⁓
understand so fully but they’re just so quick and so fast ⁓ with new AI applications. It’s like a very different way of building.
Federico Ramallo (06:40) Yes, I was having this conversation about, you know, people talk about vibe coding in the Bay Area. right? And it’s pretty much what you’re describing is ⁓ a product owner now can build a prototype using vibe coding. right? And I think that for that scenarios, ⁓ vibe coding is great. right? ⁓
Shariq Salar (06:52) Yeah.
Federico Ramallo (07:08) is never going to be the same quality as a senior engineer could build. right? But if you’re using it for disposable prototypes, that’s great. right? ⁓ What I think that the AI doesn’t understand is the abstractions and being able to plan ahead what you’re building. right? ⁓ At least not right now, right? ⁓
Shariq Salar (07:32) Hmm.
Federico Ramallo (07:36) Maybe by the time this episode airs, you know, it does. So I think that it gives the a lot of people that don’t have the experience on how to build the stuff, basically, it gives them that ability, right? And even I’ve been coding for 25, 30 years, I don’t know.
Shariq Salar (07:38) Too shit.
Federico Ramallo (08:06) For me, it also works because sometimes I have a JavaScript script that I know there is an issue there, so I can just ask the AI, can you help me fix whatever is going on here? And most of the time for syntaxes and things like that, it works great, right? Or write me a little script for doing this particular thing, right? And I don’t have to remember about which instruction or which, I can just…
use the AI for those ⁓ menial tasks. ⁓ So I think that, I agree with you that ⁓ it’s ⁓ empowering a lot of ⁓ non-technical people or with less experience to build stuff. ⁓ But I think it’s a double-edged sword, because on one hand you had that, but now you have a lot of monkeys with knives running around. ⁓
Shariq Salar (08:57) Yeah, that’s dangerous.
But do you also think, I’m curious, ⁓ as somebody who’s been coding for a long time, do you think?
It takes away the blank slate, in a sense. Do you think it ⁓ restricts creativity? Do you think in five years, if every developer is using the same models and the same platforms to code, do you think somewhere it homogenizes the process and takes away from the creativity of… It’s like if there was a canvas, and canvases came with…
set boxes and you only had to color between the lines or you just had to point and click, how many different outputs would you really get? Doesn’t that kill creativity in a sense?
Federico Ramallo (09:51) ⁓ Well, ⁓ as an expert on creativity, because I’m writing a book about creativity, right? I’m actually writing a book about creativity, you now you mentioned it, ⁓ but ⁓ I think it’s like this. ⁓ On the book what we talk about is you need movement to be creative, right? ⁓ So what happens when you have, when you’re alone with a white
Shariq Salar (10:02) Interesting.
Hmm.
Federico Ramallo (10:20) white piece of paper, a canvas, blank canvas, Most people get stuck there, right? It’s difficult to get started, right? So sometimes we have this concept of rubber duck encoding, right? Where you ask somebody for help, they don’t say anything, they just hear, right? They just listen. But that process of explaining what you’re trying to do to somebody else, that triggers like, oh, now I understand what I’m.
Shariq Salar (10:46) Right.
Federico Ramallo (10:48) trying to do because I was able to sort ideas out. ⁓ And basically, say thank you to the rubber duck, even though it didn’t provide any significant value to the conversation. So I think that ⁓ having ⁓ the way that I described AI is an eager person, an eager creative. It ⁓ never ⁓ wavers on the attitude.
Shariq Salar (10:58) Yeah.
Right.
Federico Ramallo (11:16) Even if you poke him or you can be mean with an agent and it’s always going to respond with a happy face. Or even if you ask for a stupid question, it’s going to say, ⁓ that’s an interesting take, but… And it’s going to explain why what you’re saying is wrong. But having that positive energy, that positive attitude, I think that rather than killing creativity,
Shariq Salar (11:43) Yeah.
Federico Ramallo (11:46) it allows us to be more creative, right? ⁓ Because the LLMs are average of everything and master of none. So it is up to us to say, to set the direction, to become the navigator, right? And you can ask the AI multiple questions or give me five options or five ideas. And from there you can get inspired by one.
Shariq Salar (11:57) Hmm.
Federico Ramallo (12:15) discard the others and move forward. And the AI is not going to be resentful about any of their ideas, right?
Shariq Salar (12:23) Interesting. ⁓ That’s actually quite an interesting take. ⁓ I almost seem to think that in the future, most people will have a bare coder ⁓ formula with AI. So I’m not gonna code alone, but I’m gonna sit with you. ⁓ But eventually it’s gonna be my output. And maybe it gets trained along the way on how you like to write code and how you like to solve problems.
Federico Ramallo (12:52) Yes, yes, I think that that’s where we’re going with that. ⁓ The issue that I see is that ⁓ when you’re building stuff, you’re learning at the same time. But now if you delegate the building to the AI, ⁓ then you don’t learn. So ⁓ you need to learn in order to guide the AI to where it should go. ⁓
So understanding, and that’s what I was talking about, abstraction, right? So ⁓ when you build object-oriented programming, you abstract concepts into classes, and then those classes interact with each other. So you become a designer of that interaction, right? And when you are building with AI, it kind of builds classes based on what the AI thinks, but…
It doesn’t have that concept of high level of abstraction. So I think it’s up to us to define that so the AI then can do the mechanical work of writing the code. And that’s where I think that the senior people, the people that has experience, will be able to do that better than a junior that has been relying on AI from the beginning.
Shariq Salar (13:56) Mm-hmm.
Mm-hmm.
Hmm.
Make sense, make sense. Do you also think, besides seniors and juniors within ⁓ the development ecosystem, do you think it will also make ⁓ software building available to completely non-tech folks? we both know that there are really smart people that work in medicine, in healthcare, in finance, ⁓ in investing, ⁓ and oftentimes…
There’s a context layer that’s missing between them and say development teams or between them and technologies. But if you take that away, if you take the need for that away, and if they can with no knowledge of software or code, if they can just provide the context and pair code with an AI model, do you think they will be able to build solutions that are much more effective and ⁓ long lasting?
Federico Ramallo (15:16) That’s a very interesting question. ⁓ In short, think yes. ⁓ I think that from ideation to a ⁓ shipable product of sort, ⁓ I think that that gap is becoming shorter and shorter, and AI is facilitating that. ⁓ I can tell the story of my mom used to write
code on a piece of paper, poking holes so the machine could read it. There was one machine for the whole university. So you had to allocate computing time to actually go and process your program. And then you would get a result on another piece of paper. I mean, we’re talking about lots and lots of paper. ⁓
then you will go and debug the program based on the output and then you will fix it. So the cost of debugging was high, so you will go over and over on the code. And then from that in the 70s, 80s, whatever, ⁓ to now, now that ⁓ computing, you don’t need to scale computing power because all our computers are much more powerful than they were used to be before.
Now we have more computing power than we actually need, right? That is sitting idle there, it’s cheap. So we can basically be ineffective on the computing power because who cares, right? It’s cheap and we can just use it to build things that help us solve things, right? ⁓ So yeah, that’s what I am on that, right? So if…
If a doctor can see a problem and build software with an AI agent, maybe that software is not going to be efficient, but it’s good enough for what they need,
Shariq Salar (17:21) Yeah,
yeah, think it’s right around the horizon. I think it’s already happening. I actually have to speak to a whole bunch of doctors in the Bay Area who were asking me, is AI going to do this? Is AI going to do that? I have these problems. And in the back of my head, I was just thinking, you know that day is not so far when you won’t need to speak to
a technologist about these problems, you will probably have some sort of healthcare AI or some sort of language model that just caters to you and ⁓ we’ll do that on your behalf and throw iterations and probably build you an MBP super quick and then you can test it out. think, honestly, if I think about ⁓ my grandma, for instance, probably would have used a calculator at best.
But now she knows how to use a cell phone. She knows how to text people, she knows how to call people and whatnot.
used computers at work, ⁓ but now she can do a lot more than just that. And my sister now is doing like, ⁓ you know, a PhD in data science and ⁓ you know, it’s…
People from all walks of life have so much technology integrated in their lives already that they’re slowly learning about how to interact with ⁓ technology ⁓ in much faster ways. So you have ⁓ people who previously were so disenfranchised from tech usage, like in the 70s and 80s, I would assume.
if you worked at IBM or if you worked at Xerox or if you worked at JP Morgan, yeah, maybe you use a computer. ⁓ But now, ⁓ in the developing world, for instance, you’ve got people who are selling food on the streets, ⁓ who use UPI for payments, who’ve got…
cell phones, ⁓ we have gig workers who deliver our food, who know how to use Google Maps ⁓ and this was all unthinkable technology for us, right? Like when we were kids and we were growing up we thought okay is that what the future looks like but I never thought that in the future everyone’s using technology and people we didn’t think would have access to this level of technology and now I’m wondering in the near future
Federico Ramallo (19:58) you
Shariq Salar (20:02) Will we even be ⁓ separate from technology? Will we all have, you know, like if I think about my cell phone, it’s almost like a part of me at this point, you know, I don’t exist in isolation from, you know, my personal tech devices, you know, my laptop, my cell phone, you know, probably someone’s Apple watch. These are extensions of them. So we’re already cyborgs in a sense. ⁓ And as you start integrating more,
Federico Ramallo (20:19) Right.
Shariq Salar (20:30) artificial intelligence and capabilities into those devices. I think the outcomes that we’re expecting from people are, you know, like a combination of both of those. I mean, five years ago, I could not have told you, you know, what 2025 looks like. So right now, when I think about 2030, I just pull it off. I’m just like, hey, let’s wait and watch.
Federico Ramallo (20:50) Hahaha
Yes, can. ⁓
I was thinking about movies from the eighties and the nineties and, you ⁓ where, know, they have a problem or whatever. If you think about it, you know, if, if they had a cell phone, most of those plot lines would not, you know, would not compute anymore. Right. Because it will be solved with a phone call. Right. ⁓ you know, or, you know, and I think it’s in pretty woman where, where, you know, they, ⁓ they take the concord to.
Shariq Salar (21:14) Yeah.
you
Yeah. That’s interesting.
Federico Ramallo (21:30) you know, to go from New York to London to catch up to the woman that it’s taking the regular plane and whatever, and he’s expecting on the other side. Maybe it’s a wrong movie, but you know, but now you don’t need that because you can do a phone call, right? And now you have Wi-Fi on the airplanes, right? So there’s so many stories like that that, you know, it kind of doesn’t work, right?
Shariq Salar (21:35) Yeah.
It’s interesting you brought that up because do you remember in the ⁓ in the futuristic movies of the 70s and 80s you know ⁓ if we’re thinking like Stanley Kubrick for instance right ⁓ you know there’s this there’s this approximation that in the future video calling is a thing and that it’s possible that you know you can just like look at somebody who’s somewhere far away and you can have video calls with them ⁓
But it is that point in the future when video calling is available. But people don’t even want to phone call you anymore. All people want to do is text, and the texts are getting shorter and shorter. So sometimes our visions of the future perhaps are based on some wishful thinking, and it actually does not pan out that way.
Federico Ramallo (22:37) Hahaha
Yes, yes, I agree. I was watching the Orville and this happens as well on Star Trek, right? Where they have a phone call to give them orders of, know, like, you have to do this mission and whatnot, right? And I was thinking, well, it works for the plot line, right? Because now you have the, you know, the general and the captain of the ship, you know, they’re talking about the mission and whatever. So it’s a good plot line, but, you know,
Shariq Salar (23:14) Yeah.
Federico Ramallo (23:17) that meeting could have been an email, right?
So yeah, I completely agree with that. And ⁓ I was talking with a guest recently, ⁓ Ben, and he’s doing ⁓ automation for your everyday chores. So picking up your medicine, your groceries, all that you can send your car to pick it up so you don’t have to go, right? ⁓ So you don’t need a computer science degree to do that, right? You just use your phone and…
order it, right? So I think we’re getting to that point where the functionality that this tool provides are becoming a higher and higher level of complexity than it used to be. I think that’s, you know, and lowering the, you know, the technical requirement for the users, right? You don’t need, you know, you don’t need to know computers to drive a car, you know, you won’t need to know any technical stuff to
Shariq Salar (24:05) Yeah.
Mm-hmm.
Federico Ramallo (24:22) automate your everyday groceries, right?
Shariq Salar (24:26) That’s very interesting. In fact, ⁓ I’m actually looking forward to seeing what this is going to do for healthcare and drug discovery and pharmaceuticals and anti-aging. think those are really, life sciences I think is like the most interesting space in which I think there will be applications of AI. I don’t think so much of it has been adopted.
because the R &D cycles are really long and it’s generally deep tech. But I think in the near future, we want to see a lot more of that happening. I’m super excited to see what’s going to happen there. Are there any spaces that you’re super excited about with AI?
Federico Ramallo (25:09) Yes, so I think that AI is becoming better and better on a higher level of abstraction for problem solving. ⁓ So I’m looking forward to see how that’s going to improve the quality of the code, right? ⁓ Where eventually we might not need to code anymore, right? Or it’s going to be, I mean, getting to that ⁓ shift in paradigm. I’m looking forward to that, right? ⁓ Where you can…
Shariq Salar (25:19) Mm-hmm.
Federico Ramallo (25:38) describe functionally what are your expectations of behavior and then the application is going to… the LLM is going to build the application as you need it,
which is going to leave me out of work as a software engineer. But that’s a future me problem, right? Yeah.
Shariq Salar (25:51) Interesting Yeah, I do you
know, honestly, I don’t think it’s just software engineers who thinking about this right now It’s also people in in governance and policymaking and economics. you know There’s all these people who read, you know a thousand big books to be where they are and you know ⁓ Spent, you know 10,000 hours to you know, like
really learn a subject and now here’s like this little box, this, you know, like that, you know, probably fits on a pendrive and it’s probably going to give you better analysis and better decisions. and better, you know, it’s just a question of how long you take to fine tune it and give it enough context. And I think everybody in so many different spaces is actually thinking how is this going to affect
You know, I’m just like, if I just glance around like LinkedIn or different forums, graphic designers you see talking about this, animators are talking about this, filmmakers, know, most people in creative professions are definitely talking about. So traditionally when we thought about subjects in the humanities, like, you know, like the creative professions, ⁓ you know, since you mentioned it earlier, generally they were… ⁓
affected by you know big advances in technology. ⁓ If something is made it’s made for their use and it’s you know made around you know their current needs and you know their processes but ⁓ now the kind of you know things that we’re building in AI for instance if you look at you know the most recent ⁓ you know models ⁓ text to video or text to image or you know text to voice and ⁓
those models. ⁓ It bypasses so much. It’s like how you said that meeting could have been a phone call or that plot line is completely irrelevant. ⁓ Ideating is now completely irrelevant. You can in the future serve 10 different endings of a of film or a TV series to 10 different viewers.
I had a friend who just recently ⁓ finished a book and a series of books and she said, can you build me some AI to like write like, you know, four more, you know, editions of this? Cause I don’t want the story to end. And honestly, I said, hey, that’s actually a very possible use case, you know? ⁓ I’m just hearing so many of these like on a daily basis here.
Federico Ramallo (28:40) We we’ve done something like that with my son. We were on the car waiting for my wife I don’t remember we were like hanging around right and kind of killing time and you know kids get relentless in the car really I don’t want to do anything I want to do something so I I Opened up chat GPT and I said, please tell me a story of he likes ⁓
Shariq Salar (28:44) Okay.
Federico Ramallo (29:07) this Paw Patrol, know, these little dogs that go and save the day, right? Anyway, and then I described the story that I want and I included my son into the story and then I included my dog into the story, right? And then I said, ask me questions so we can continue the story, right? And it will give us like a little, you know, initial story and then we’ll say, oh, and now it’s a dog or now
we’re going to this place or now, you and then in a few minutes, Mason was so engaged with that, you know, and he was building the story with the AI, right? And now, you know, a space rocket shows up and, know, aliens shows up and, you know, he was, you know, he started coming up with these different ideas and it was so fun, you know, so much fun.
Shariq Salar (29:38) Hmm.
That is so interesting. I think you know, you know how ⁓
Millennials and Gen Z were the first to have a very strong dependence on, let’s say, search engines and Google and say Wikipedia. ⁓ my generation, think.
We needed to learn things and remember things and the generation after me, because like, you know, I see so many of my colleagues, they don’t think they need to remember anything because they were like born Google native. So when they when they were like, you know, in their teenage years or their 20s, they were just like, yeah, I could just Google that. I don’t really need to know much. ⁓ I think your son might be from the generation that is the first AI native generation. So it’d be interesting to see, like in 20 years, you know, how they think about
Because it’s been around since they were a kid.
Federico Ramallo (30:59) Right.
Shariq Salar (30:59) it’ll be interesting to see what they do with it. ⁓
Federico Ramallo (31:03) Yeah, I I think you probably remember before Google Maps, right? We had to use maps, real maps to actually go places, right? And now with Google Maps, we kind of become dependent of it, right? Like, I wouldn’t know how to go to many places without Google Maps, right? How many phone numbers you remember by heart, right? ⁓ I mean, I used to.
Shariq Salar (31:09) Yeah
Yeah.
for sure. I wouldn’t go around the block without looking at the map once.
Zero.
Federico Ramallo (31:32) But now,
I am dependent on the phone, right? So yeah, I think that kind of shift is going to happen. Whether we like it or not is going to happen.
Shariq Salar (31:45) But what do you think it frees up memory for? So if you know there’s a block of ⁓ brain power that was allocated to storing things, and now that doesn’t need to happen anymore, do you think that frees up space for more intellect? Or do you think that frees up space for more empathy?
Federico Ramallo (32:08) You are thinking like Sherlock Holmes. You know that story, right? So Sherlock Holmes ⁓ described to Watson that he only remembered what he thinks is important and he discards everything else so he can have more space in his brain to remember things that are important to him, right?
Shariq Salar (32:12) Hahaha.
No, please, go ahead.
Federico Ramallo (32:38) So he would remember very, I would say strange, but non-traditional things that would help serve a purpose to solve the mysteries of the murder or whatever. So he would remember crazy things like the color of a dirt from a specific region and things like that.
Shariq Salar (32:38) Mm-hmm.
Yeah.
Federico Ramallo (33:07) but he will not remember the name of one particular person or something that most people would. So it’s that concept of what you’re saying. I think that the… I mean, there is some truth into that, but I think that the brain is malleable. So if we do need to remember things and we push ourselves, there is no limit to what we can do. ⁓ Do or learn. ⁓
Shariq Salar (33:34) I agree, yeah.
Federico Ramallo (33:37) So, but it does require a an effort, a conscious effort towards that that that place, right? ⁓ That that that remind you go ahead.
Shariq Salar (33:48) I do agree, I do agree. I do hope that,
sorry, I was gonna say I do hope that as a global set of people or as a race, we have some clear goals that we all want
positive goals that technology can help us realize. I just hope that with something so powerful, I really hope that in 20 years when we look back, I just hope that we made the best use of it. Yeah.
Federico Ramallo (34:29) Yes, I hope the same. my My concern is that there’s so many interests that it’s hard to find this consensus towards a goal that helps humanity. ⁓ I like the concept of Star Trek or these other sci-fi movies where they say, ⁓ we went beyond this ⁓ concept of…
resource holding right ⁓ money is no longer ⁓ we don’t need money anymore right which sounds great on the movie but I don’t know how that’s going to be possible to do in a you know in reality. right so that that yes yes I I love Star Trek
Shariq Salar (35:05) Yeah.
Are you a Star Trek fan? ⁓
So I’m
not really a Star Trek fan. I don’t know enough about it. But since you’ve mentioned it twice, I should tell you that Atrean is actually a type of character in the Star Trek franchise. So Atreans are a particular species that have deep listening abilities. I guess if you Google Atrean, you’ll see.
Federico Ramallo (35:31) really?
I’m doing that. from the original series. Interesting. The lizard people.
Shariq Salar (35:45) Yeah.
you
Federico Ramallo (35:52) Interesting. Now I will have to look it up. Yeah. I have to read more about this. Yeah. Interesting. So.
Shariq Salar (35:58) That was just a coincidence,
But you know, ⁓ it’s interesting how many of our current technological advancements are actually based on science fiction or futuristic TV series and whatnot from our childhood. ⁓ I actually didn’t think I would be seeing so many of those things realized ⁓ in my lifetime, and here they are.
Federico Ramallo (36:33) Yeah, I I think ⁓ that it’s inspiring. Science fiction from the 50s, that’s kind of where started. They had these crazy dreams about what the world could become, right? And, you know, by this time we should have flying cars, right, according to science fiction, right? Yeah. So I found that very interesting.
Shariq Salar (36:43) Hmm.
Yeah.
Federico Ramallo (37:02) It’s very inspiring, actually. And even though sometimes they’re completely wrong, sometimes they hit it, right? It’s a hit and miss, but it is inspiring nevertheless, right? And the person that I think talks more about, I need to look up the name of the author because I never remember.
Shariq Salar (37:05) Yeah.
some off.
Federico Ramallo (37:32) Asimov, Asimov, yes. He’s the one that talks a lot more about this, I want to say dilemma, but struggle between humans and technology, right? Which is kind of where you’re going with your questions about how AI is going to change our lives, right? ⁓ He even has, ⁓ on one of his books, he talks about this concept of
Shariq Salar (37:33) Isaac has more.
Right.
Yeah.
Federico Ramallo (38:00) this eternal being that, know, human that becomes eternal by changing part of his body. And then eventually he gets to a point, I’m spoiler alert, you know, ⁓ he, he, he was learning with, with neck future generations of humans, right? Until he got to a point where, ⁓ everybody, everybody else got, ⁓ the new thing, right? Understood like, you know, the new thing. And he was,
Shariq Salar (38:11) Hehehe.
Okay.
Federico Ramallo (38:30) unable to understand that. And that’s the moment when he realized that he became obsolete, that he was unable to learn new things. And he was unable ⁓ to keep up with the learning of society. He’s been ⁓ helping society and humanity ⁓ to improve because he was at the edge of the technology, because he was
Shariq Salar (38:40) Mm-hmm. ⁓
Federico Ramallo (39:00) a scientist and he was, you know, ⁓ he knew a lot of about a lot of things and he was, he had so many lives. He has lived so many lives that he has so many experience until he got to that point. And that’s kind of what he decided. Okay. I, I, I, this is the time where I, I, doesn’t make sense for me to keep living because I become obsolete for humanity. Right. Yeah. I’m probably butchering the story and paraphrasing, but
Shariq Salar (39:23) Mm-hmm.
No,
Federico Ramallo (39:28) That’s kind of what I remember. Yeah, yeah, yeah. So yeah.
Shariq Salar (39:29) that was really good. But you know what I was thinking through, what I was thinking through, you know, your narration was in in the 50s, if there was a normal, sane person and he was thinking about cars flying someday, that by definition is a hallucination, okay?
Federico Ramallo (39:57) hahahaha
Shariq Salar (39:57) Because
it’s not real and there’s no way to make it real in the present day. But it’s an aberration in thought, right? And I’m thinking, ⁓ will we have models whose job it is just to hallucinate reasonably to try and help us predict what the future looks like? ⁓
because at the end of the day, it’s the job of large language models to have some hallucination. And that’s how creativity happens. That’s how new words or new sentences or whatever are made up. Perhaps if imagination or hallucination was the key to ⁓ imagining a future and creating ⁓
ways for us to build. How much more of that can happen using AI? I’d be interested in seeing that.
Federico Ramallo (41:06) haven’t thought about it that way. That’s very interesting. When I implement, use AI tools and implement them on workflows, I am boxing, I don’t know if boxing is the word, putting, trying to mold the AI to become deterministic, even though it’s a non-deterministic thing, right?
Shariq Salar (41:32) Exactly.
Federico Ramallo (41:36) And now that you’re talking about that and realizing, why do we have to fight with this so much? It’s because it’s non-deterministic. What happens if we encourage that, right? Wow.
That would be, yes, that would be something interesting.
Shariq Salar (41:55) See, I think this is something for the next generation to do, because I think usually we would be very cautious with the new technology, as is like, it’s a very human thing to do that. You when you see something new and super powerful, like, hey, we found fire, we’re gonna be like super careful with that. But the next generation, they’re gonna try and figure out how to put it on a matchstick and you know, how to, you know, do pyrotechnics and whatnot. I really think, say for instance, okay,
Three years ago, what was the one thing that everybody in the world was thinking about, day and night? It was a virus, right? It was a pandemic and it was the COVID virus. And you could not have gone through a single day without thinking about it. Now, let’s say hypothetically,
10,000 years ago, a catastrophic event like this would happen once in 10,000 years. Then the incidence rate of that increases to once in 1,000 years. And then the incidence rate increases to once in 100 years. And then it goes to once in 10 years. And now, ⁓ what if that accelerates to a pace where it happens once every year?
It would be highly unfortunate, but I think if…
microbes, organisms, pathogens, can become superbugs and super intelligent bugs much faster. We will need to create a super intelligence that helps us figure out how to deal with it also much faster. You know, when the pandemic happened,
As a race, we were out of options, we were out of solutions. We didn’t know how to counter that or control that or treat that or manage that. ⁓ I think one of the best things that might come out of more advanced AI with super, super computing powers may be ⁓ dealing with
existential threats to humanity because we’re just like at the cusp of how much we can do with our own minds and how much we can do collectively, you know, as people who put their minds together. But ⁓ beyond this, I think, you know, the problems of the next generation will require very advanced forms of intelligence. ⁓ And so we just pass that threshold. So I think that could be another thing that would become super useful in the near future.
We have no choice, we have to build it.
Federico Ramallo (44:42) Right, right, we have to build it. mean, it becomes a survival situation where either we build it or we perish, yes, as a humanity, right? ⁓ There’s also something interesting in what you were describing that, I mean, on all of these ⁓ crisis, right, or on all of the situations like the pandemic.
Shariq Salar (44:52) Yeah, I guess, yeah.
Federico Ramallo (45:10) There were a lot of people that died, which is unfortunate, but also means that there is a paradigm shift in humanity, ⁓ which is kind of what happens with cellular regeneration. Now you have ⁓ new cells doing new things, different than what we used to do before. So ⁓ I think that’s going to happen with us as well. ⁓
our rate of change is going ⁓ to generate changes that we couldn’t see coming now,
It’s going to be very interesting world.
Shariq Salar (45:50) I agree, I totally agree.
It is, it is. I hope that we live long enough to see it. But I think it’s getting, it’s looking very exciting for the next generation. I just hope that, you know, we leave a world ⁓ good enough for them to feel that excitement and to, you know, enjoy it and thrive in it.
I think, you know, when we think about how much of current advances in technology are focused on humans and human comfort and human abilities and human, you know, longevity, I would like perhaps for some parts of, you know, at least 20 to 30 % of that, you know, effort to also be dedicated to
the rest of our environment surviving and thriving and to be enjoyable in the future. Unfortunately, I’m not seeing so much of that right now, but I personally am of the belief that if the Earth is lost, everything is lost, right? I would certainly continue to want to see us as a race on this planet ⁓ living in much better ways.
like to see maybe like a little more you know ⁓ tech focus on ⁓ not just our internal systems and us but you know what’s around us in how to preserve that and how to make that better I think that would be interesting to see yeah
Federico Ramallo (47:29) Yes, yes, think that that would be very interesting to see coming. ⁓ My concern is the, you know, like the other side of the coin, ⁓ Or an alternative vision of what you’re describing is, you know, before the 2010s, nobody has, the kids didn’t have any phones, right? So the level of ⁓ depression and disconnect that people had.
Shariq Salar (47:51) Mm-hmm.
Federico Ramallo (47:58) that that kid had ⁓ has been increasing since they had cell phones and social networks, right? Because social networks allows people to be mean to each other, right? ⁓ Without the consequence of, you know, physical being presence. ⁓ So…
that generated a generation of kids that became more depressed, right? And less connected with each other right now. Everybody’s on their phones, right? We feel that we have more communication that should connect us more, but in reality is disconnecting us more, right? ⁓ The level of birth rates is going down, right? Or what happened in Japan that most kids at 30,
not kids anymore right but people at 30 they are ⁓ alone they don’t have a partner right ⁓ so they had issues finding partners right ⁓ so my concern is that ⁓ the AI will take you know forward to that direction right why would a kid ⁓ prefer to date a woman when the kid can date
an AI, you know, virtual whatever, virtual woman that it can be ⁓ molded to be perfect as the kid wants, right? ⁓ Where if you date a woman, a woman is going to be less perfect, right? Or imperfect, right? So I’m wondering if we’re going towards that direction, right?
no, we lost Shurik