This Week in Startups - Magic.dev CEO Eric Steinberger on making developers bionic | E1744

Episode Date: May 17, 2023

This Week in Startups is presented by: Crowdbotics. Great ideas can change the world, and Crowdbotics is the fastest way to turn those ideas into code. Get a free scoping session for your next big app idea at http://crowdbotics.com/twist. The Microsoft for Startups Founders Hub helps all founders build a better startup, at a lower cost, from day one. Startups get up to $150K in Azure credits, access to free OpenAI credits, free dev tools like GitHub, technical advisory, access to mentors and experts, and so much more. There is no funding requirement, and it only takes minutes to join. Sign up today at http://aka.ms/thisweekinstartups. Release. Large enterprises pose unique challenges for SaaS startups. Unlock customers with unique needs for private and single-tenant hosting without the toil of DIY with Release Delivery. Get your first month free at https://release.com/twist. * Today’s show: Jason is joined by Magic.dev CEO and Co-Founder Eric Steinberger to discuss how his startup makes developers more efficient with AI by auto-generating code. The first portion of the interview features Eric demoing his software. Then, Jason and Eric dive into the impact that these rapid AI advancements will have on society (31:28), how their tools have the potential to significantly minimize coding time and what that means for entrepreneurship (13:55), and much more! * Time stamps: (00:00) Eric Steinberger joins Jason (1:44) Eric and Jason explain what Magi.dev is (3:43) Eric Demos Magic.dev (9:13) What makes this AI model unique (11:02) Crowdbotics - Get a free scoping session for your next big app idea at ⁠http://crowdbotics.com/twist⁠ (12:11) Magic.dev's motivation behind building their own language model (13:55) The efficiency gains achieved with Magic.dev (17:32) Exploring reinforcement learning and grounding the AI model (19:45) Microsoft for Startups Founders Hub - Apply in 5 minutes for six figures in discounts at http://aka.ms/thisweekinstartups (21:19) The two parallel paths unfolding (23:57) How close we are to developing more capable AI systems (31:28) The impact of Moore's Law on technology and society (37:34) Release - Get your first month free at ⁠https://release.com/twist (39:03) Alternative perspectives on the pace of AI advancement (43:14) The origins of Eric's positive outlook on AI (47:49) Thought's on AI regulation and choke points * Read LAUNCH Fund 4 Deal Memo & Apply for Funding Buy ANGEL Great recent interviews: Brian Chesky, Aaron Levie, Sophia Amoruso, Reid Hoffman, Frank Slootman, Billy McFarland, PrayingForExits, Jenny Lefcourt Check out Jason’s suite of newsletters: https://substack.com/@calacanis * Follow Jason: Twitter: https://twitter.com/jason Instagram: https://www.instagram.com/jason LinkedIn: https://www.linkedin.com/in/jasoncalacanis * Follow TWiST: Substack: https://twistartups.substack.com Twitter: https://twitter.com/TWiStartups YouTube: https://www.youtube.com/thisweekin * Subscribe to the Founder University Podcast: https://www.founder.university/podcast

But people should just take note. I mean, the fact that, you know, people who are building this stuff, Jeffrey Hinton, the guy who's just a godfather of AI, of course, but he drove this, that he's worrying of it. And everyone, I mean, I, you know, I'm just a tiny fish in the pond. The people who've been working on this for decades are worried. This is different from when people are worried about, you know, crypto, like, or when people are worried about whatever else is the thing they're currently worried about, right?

That the people who have been working on this for ages, who have dedicated their careers to it, including myself, are saying that this is very much a path with a juncture. This weekend startups is brought to you by CrowdBautics. Great ideas can change the world. And CrowdBotics is the fastest way to turn those ideas into code. Get a free scoping session for your next big app idea at crowdbotics.com slash twist. The Microsoft for Startups Founders Hub helps all founders. Build a better startup at a lower cost from day one. Startups get up to $150,000 in Azure credits,

access to free OpenAI credits, free dev tools like GitHub, technical advisory, access to mentors and experts, and so much more. There is no funding requirement, and it only takes minutes to join. Sign up today at a.k.a.m.s.

This week in startups. And release. Large enterprises pose unique challenges for SaaS startups. Unlock customers with unique needs for private and single-tenant hosting without the toil of DIY with release delivery. Get your first month free at release.com slash twist. Hey, everybody. Welcome back to the program today. We have another great guest. We're doing this like AI series where we're interviewing anybody and everybody who's making interesting stuff, builders in fact.

of the most interesting AI startups. Today, we have Eric Steinberger. He runs a company called Magic. And plain English, Magic is building a colleague inside of a computer. Basically, it's powered by AI, and it's going to auto-generate code based on a developer's inputs.

Welcome to the program, Eric. Hey, thanks Jason. Thanks for having me. All right. How many people work at this company? We're 10 now. 10 people were you based? Half of us in SF, half of us in Vienna in Austria.

Okay, great. And you're in SF? I will, I'm just moving. I'm going to my visa, have my visa appointment in one week. Well, congratulations. Are you going to be in San Francisco or are you going to be in the Bay Area? Yeah, I plan to be an SF. Wow.

Would you have an office there? Yeah, yeah, small office, you know, small team. Whereabouts in town are you? Not sure yet. We're probably going to get a co-working space just to be quick, for the first phase and then eventually. The reason I ask is a lot of AI people are saying they're going to be in the city of San Francisco, where a lot of startups and obviously the big tech companies are no longer going to San Francisco.

Explain the decision to be in San Francisco as opposed to saying being down in Palo Alto, Mountain View, Cupertino, San Jose, San Mateo, wherever. I mean, I have a few of friends who are just really good. I'm L engineers NSF. And I'm not very familiar with the area. So I think NSF is the place that I've been. I've spent some time to Berkeley, but I think SF is better for what we're doing. Great.

Awesome. It does seem like a bunch of AI developers are choosing to be in San Francisco, which is great for the city since there's like 15 Salesforce towers of open space. So let's do a demo here. It's always great to demo. But when we do our demos here on This Week in startups, we sports cast them, which means we try to explain what we're seeing on the screen. So let's pop up your screen and let's show people.

We like to show rather than tell. And in this case, we'll do a little bit of both. So if you're just listening, we'll describe what we're seeing on the screen. So Eric, take it away. Great. Thanks, Jason. So, you know, after spending about a year in the cave, building an LLM training and inference framework

to build our own code models and scraping a lot of code data, we finally managed to build something that we think has a qualitative edge on other products in the space. you're familiar with code completion tools. They give you completions. You know, I can like, I'm going to go put a new line here, and I'll get a completion.

That's standard. So what you're showing on the screen is you're developing in Python maybe? I can't see. Yeah, that's right. I'm doing, I'm developing a writing code. I'm writing code in a Python project. Here, this is like a reinforcement learning code base.

I used to do research in a few years ago. Great. I'm in it. So when you start typing in a command? They're like, you know, so if this is like, it's, it looks like co-pilot on first glance. I got like, you know, I get, like, you know, I get completions. And so, I should say this model, we're just training, where we're still training it, it's like sort of an intermediate checkpoint.

But the one thing I want to demo here is that the key difference, you know, copilot takes in, like, say, the current file or maybe, you know, like a few adjacent pieces of information. And you're referring to GitHub's, you're referring to GitHub's, co-pilot. which is the original OG auto completion. So if you were using Gmail as a user and you see it shows you one word and then it started showing two or three words, this is something that developers use. And developers have a more constrained possibility set than, say, people just writing email. Is that correct?

So therefore, it can maybe give a little bit more code and it knows the existing code base, right? So co-pilot doesn't know the existing code base. And I think that's the key thing. That's exactly what I wanted to allude to. You're right that coders, you know, a lot of code is predictable, right? Sometimes it's just like super clear what the next line is. Sometimes it's really hard. And co-pilot generally and general, like co-completion, auto-complition tools,

find it easy to complete things that are straightforward and somewhat obvious. And it's quite harder to do hard things, but, you know, kind of getting better at those with language models being scaled up. But they have this limit of context size. Even GPT4, you know, it can fit more, but it's still limited. What we've developed over the last year is a technology that can really look at all of your files. It can see your whole repository. So you can see here, I've opened up a few folders just to show that there are a lot of files. And I can keep going.

Like, the folder depth is pretty, you know, substantial. This codebase overall is, I don't know the number of lines or characters, but much, much, much, much larger than any of the GPs could fit into context window. But our model does. So for instance, if you were to auto-complete from the start of a file, usually here, if I were to go to the very first position, the model has no idea what I would be about to do unless it sees the repository. It's starting from a blank canvas.

So if I take out here, for example, if I take out my name and I want to type something, okay, just demos. If I start typing, it knows that I wrote the code base. And how does it know that? Because it goes to other files where I have written this same line. But at the same time, you know, you can do your normal completions. So, you know, as you said, like this is, for example, a more obvious line, right?

Like here, I have something similar above and then it adds this minus one piece, which is intelligent because it's like player zero, player one. It figures out that I want to access the other player. So it just sort of, as you see, it just told to complete that for me. Okay, so when you say that, what you mean is it's you have a string of code above it. It recognizes that code. Your learning model knows and your auto completion model knows the code you wrote above it. It assumes now that you're writing about another similar thing and it can then put in gray, hey, here's what I think you're going to do next.

It reads your mind. And then instead of typing, you just hit like the tab key or something to fill that in. Exactly. That works great. So just like in Gmail, but this is as if it had read every one of your Gmail's. And it knew that you like to often, let's say if you were a VC, you might talk about your portfolio companies, you might talk about other partners in your firm. It would have that information.

So when you says, hey, you should talk to my partner, Susan. It fills in a little bit about Susan. Exactly. It's able to take in the context of what the developer has done and produce code that is similar. As you said, I think you said this very, if you put it very well, it reads your mind and tries to do what you were about to do. And you know, we can do a few more like this, right? I can write this line of code. And because the function is called do pickle, which in Python just means store a file in a certain format to disk, it just wrote that for me, which is of course relatively straightforward, but it's very useful to do that faster.

But now the thing that, again, our AI does different from existing ones, we just wrote this function, just do pickle function. If I go very, like, completely different file, where here, where I use this function, and I can even close this, right? We're not using the fact that you just edited this. It just sees the entire repo. So if I, this should hopefully work. This great text, the Intelligent suggestion knows it

because it's semantically analyzing the code, which is if it goes down at free of import statements and so on, very specific, doesn't always work, only works when it's kind of syntactically clear. But here our model successfully understands that we have a method called do pickle somewhere in a repository, and that's probably what we want to use. If you tried this with co-pilot, for instance, GitHub co-pilot or other products, you would find that because it is limited to the current file and slightly small bits of adjacent information

that are only grabbed heuristically as in algorithmically designed and not learned by the model, it would find this difficult. And so what we find in practice is that as you code, when you write code, you know, you're often have a lot of code already in your repository, and our AI is able to reuse your code, to use your coding styles, to use functions from across your repository in various contexts across the project. And so that's just what we've been working on over the past months or past year at this point, and working on a few other really interesting things, but very excited to get tangible models now and launch very soon. We're giving you a sneak peek here. The model will

finish training relatively soon and then planning to launch that with the same device I just show you. When do you think you're going to launch it? Yeah, this month. Great. Probably the most common challenge I hear from founders is related to building. Either they are in technical and are searching for a technical go founder or they can code, but they're just spread to then.

This is one of the first major obstacles you're going to face and I know how discouraging it can be. But there is a solution. Do you have a great idea, but you don't have a technical co-founder? Well, CrowdBotics can be your CTO as a service. Boom, just like that. This means you can focus on building an awesome product and delighting your customers, rather than wasting your time on infrastructure planning, architecture,

compliance, and all that boring stuff. CrowdBotics also offers professional scoping to help you flesh out your project at the MVP stage and beyond. So cut out the hassle and get back to building that perfect product for your delighted customers. When you think CrowdBotics, I want you to think getting your time back to focus on product. Product drives everything in a startup. So let the folks at CrowdBotics show you how it works. Schedule a free scoping session and get your detailed build plan at CrowdBotics.com slash Twist. That's CrowdBOTICS.com slash Twist.

And so when you have a company like this, you have to build your own language model. or do you use, say, chat GPT's existing language model, which has been trained on a lot of open source products, and then do a local version of the repo of this company's code, and then merge those things together, or are you building also your own, so you're not dependent on chat GPT4, or open AI for that matter?

Yeah. We built our whole language model training and inference stack from scratch. The main reason for that is that we wanted to be able to innovate on the architecture of the model. We do a few things different from the sort of the standard way, which is also why it's taken us so long to launch something. The way we really look at CodeGen and the CodeGen space more broadly is that, I think ultimately there will be a user experience that frames the idea of fully automated

development in a way that anyone and everyone finds attractive and useful and reliable enough. There will be some way to talk to your computer to automatically write code. And that UX will replace whatever comes before, of course, including auto-completion, because if you don't write code in an editor, why would you need auto-complition? So we're very focused on making sure that what we do in the short term and in the mid-term and in the long-term is geared mostly towards the final stage of what this space will look like. And that's optimized for the immediate. We could have built something like what I just showed you within a month of just existing as a company. but if we just took other people's models and other people's inference servers and so on

and those things would have worked that they would have been interesting products and then would have probably brought in some revenue but it would have slowed us down on that So you're starting with a tool for developers so that they can write code faster. Hopefully they can write code 50% twice as fast.

I don't know what you've... What is the benchmark you think you'll be able to achieve this year? It depends on how you measure it, right? like you, there are some types of engineering that are inherently dependent on contextual understanding outside of your codebase, where you just need to talk to people. I think those numbers are often not,

like those things are often not considered an estimate. Then there is some coding that is really just boring template writing or we hope that it just does it all for you. It really depends on what type of work you do. But, you know, like speedups and this space tend to be in the range. Yeah, yeah, it'll be. Just take the average group of developers

You got 100 developers in a development team at some mid-sized startup. What would you expect across the range of developers? The average would be each year in terms of gains from this type of technology yours and other people's. Yeah, I think if you measured the time it takes a group of people to build a piece of software, as is with and without the technology, which I think would be like the best benchmark

to really quantify productivity. Sure. I think it is possible to get beyond the 2x mark, even with a few more steps of innovation. I think right now this number is below the 2X mark with existing tools like others out there. So you think the current tools are making people 20, 30% faster? I don't want to invent numbers now,

but the numbers published by various companies use different metrics. And I think ultimately if you measure it end-to-end, how long does it take you to build a production scale project time? There's just so much work in there that's not writing code. And I think that's also something we're really optimizing for it. Like, a lot of code gen companies are like, okay, let's automate writing unit tests. But how much time do you really spend writing unit tests a bit?

But, you know, not, that's not your smartest engineers. And when you say you think people will get twice as fast, do you think that's going to, that would be like every year they get twice as fast or ultimately they'll be twice as fast? How do you, what's the time frame in which you think if we were to come up with our own Moore's law, and I forced you to bet, the time it takes with AI for the average developers, become twice as efficient, all in. They can basically finish a project

in half the amount of time. How many months will the doubling take on average going forward? I think the first AI doubling should be done this year if it didn't already, if it didn't already happen for some developers,

then that should really be something that happens this year. So how many months do you think it will take on average? I think it will be much faster than doubling every constant number of months So yes, at some point there is, there's this inflection point where AI is reliable enough to just write code, make a pull request.

You'd trust that it's right. You know, the same way, we have great engineers. If one writes a pull request and another one reviews it, I don't even look at the code, right? It just goes in. And there'll be a moment when AI writes a pull request and reviews its own pull request, and it's as reliable or more reliable than your engineers. Wow. And a difference between, that isn't now, right, but that will come.

And the difference between sort of things like self-driving cars and that is that a company can take a look at this and go like, well, this is clearly better. no lives at risk, right? Let's go do it. Got it. So you think that's, it would be faster than Moore's law. I think it looks like this.

Got it. And so what you're drawing is a slow, sloping up curve and then a huge giant exponential spike. Yeah, no, this is, I think, I would agree with your central tenant. And the key issue is, can you trust the code?

And right now, you can trust the co-pilot because every line, Every suggestion it's making, a human decides yes or no. But correct me if I'm wrong, if enough humans are saying yes or no, that's reinforcement learning. The model will look at a correction that's made. If it gives a bad suggestion, it says, no, I don't, that's not what I'm looking for in terms of the auto-complete. This is what I'm looking for.

Does it take that into account the next time it's going to do an auto-complete? So I think reinforcement learning plays a key role and all of the but the reliability piece is bigger than that. Because, so this is actually, I love talking about this. Thanks for bringing it up. No matter how large you make a model, if it's not aware of its own actions of in the past, in a given interaction, it is just physically unable to remain grounded in its history. So to be more concrete, I suppose, imagine you forgot what you did before we

we started recording this podcast, and I asked you how your day was. All you could do is invent something. You wouldn't know better, right? And so a lot of the issues we see with hallucination and truthfulness and reliability, you could think of as a branch of those things comes most likely, I think, from the lack of grounding in the models past. It is sort of the average, like the surface area of the training set were just like a cloud of points, but the surface area doesn't perfect.

represent the points. And I think if models were to perfectly represent the points, they would solve a lot of these reliability and hallucination issues. So there's there are parts to it that aren't RL, but RL plays a key role. And I'm certainly, I certainly believe strongly that you need it to get there. But it's, it's so far an open question how you make models as reliable as humans. I think the key there is to get them to know that, to get them to say that they're unsure when they are unsure, instead of just trying and then like acting like they're confident. All right, everybody. Our friends from Microsoft are here.

Tom Davis, a senior director at Microsoft for startups. And you're a former founder. You are here today to talk to us about the giant leaps that Microsoft has made in the AI space. You've been giving Azure credits to startups. And that's delightful and amazing. But people really want access to the Open AI API. Yeah, absolutely. So there's two things. First of all, we've got a benefit that we offer our

startups. They can get $2,500 worth of Open AI credits. So they can get access to the latest and greatest models that Open AI are delivering. But then they get access through the up to $150,000 worth of credits that we offer through Founders Hub to leverage the Azure OpenAI service, which has a full SLA around it. So when they want to go into production and really have that reliability that we provide with the Azure SLA, they can leverage the Azure OpenAI service APIs. And they can do things like the GPT models with codex for the coding and also for the Dali models as well for images. So it's a full service.

It's not just the great APIs that you get and access to the LLMs. They can build out their own LLMs using open source. And then they can manage those with our AI tooling services as well. Amazing. Well done. And if anybody wants to sign up for that, do it now. while you are in front of your computer, AKA.m.s slash this week in startups,

aka.m.S. slash this week in startups. Well done Microsoft and well done, Tom. So there's two paths that are occurring here, which you alluded to. One is you're making developers more efficient, more reliable at their job, but then there's this other path, which is, hey, can I just talk to the AI

as a non-developer? Maybe I'm a product manager or the CEO founder of a company, but I'm a non-developer founder. Or I'm just making something small and I say, hey, listen, I want to make an app that does a high interval training and I want it to be able to track my GPS,

log in with my Google login and then put the results into a Google sheet so that I can see them there. And I can see my sprints and my recovery time and please put my heart rate in there. It could write that code and we would trust it enough and have it published to the app store or whatever to test flight,

and you wouldn't need a human involved. Those are the two paths that humanity is on right now, correct? Yes. You're choosing to build this tool, the auto completion, with the civilian, the non-developer in mind. Is that correct? Is that my understanding of your startup, magic? I think as a company, we're not just one product.

we're building towards AGI and safe AGI in particular. The code completion product we built is a very good way for us to have early grounding in user experience and customer feedback and, you know, potential early revenue and so on. But we're thinking like a lot further than just code completion. I do believe that in the near term, this product will be the primary driver of our interaction with the outside world. And I mean, you know, we, as we're nearing the launch, you know, we're talking to folks who, you know, like to use it internally.

And I should say one of the key differentiators here is that we want to offer a much more privacy-focused version of this experience in a sense that, you know, companies that don't want to run on a public API that don't want to send their code to Microsoft, OpenAI, or others. we offer private cloud solutions where enterprises are able to host their own version of our system. And so we think a lot on this in the early days now, we think a lot on the axis of product differentiation

and just early deployment of our AI models. But ultimately, I would say we are mostly focused on building systems that are much more reliable that you trust with much larger tasks where you'd instead of going and asking for like two lines, you ask for a day worth of work and give it, you know, three minutes time and then it makes that...

How close are we to that? You know, we'll see. We're working hard. Well, but just generally the industry, because right now it seems like doing snippets pretty easy. Sometimes it's going to pull in you know, some elements that, you know, might be reused over and over again, libraries, let's say. But then putting together multiple pieces to build the app, as I described it, you know, that this magic moment where some startup founder or just some civilian on the weekend wants to build an app and they can just describe what they want and have something come back to them because we see people doing that with images right now.

We see people doing that with text. hey, I want to write a blog post, and I want it to be about how to utilize social media to build a company, and it gives you a lot of great ideas, and you have to polish it off. People seem to think that those blog posts will be ready in the next year or two to be published without needing polish, right? So I guess that becomes the issue here is the last 10% of self-driving cars, the last 10% of robotic kitchens, it just gets really hard. So if you had to think of a year when people could do the task I mention, what year would it be? I think it's certainly possible that it's further away and that we'll bump into the same 90-10 type problem as you described, you know, the self-driving car world and many others.

I agree that there aren't that many applications of technology automation where you don't run into the final reliability step. But ultimately, I mean, what people are building here is a system that is converging to intelligence greater than those of humans. And ultimately, I think it is whether that's good or bad. It's a different question, but I do think it is feasible that that happens in the near term. Near term being under five years. Yeah. Yeah, I'd be comfortable saying that. Okay.

Because, I mean, I think that's one of the fascinating things about self-driving. say we've been two years, three years away for about 10 years. You know, in my estimation, and listen, I use full self-driving. It is amazing. But you must keep your eyes on the road, hands on the wheel, because there'll be a left turn or there'll be a bicycle that comes out of nowhere. There'll be, it's not ready for it to take the steering wheel out. Although Cruise, I guess, has some on the road in a very short, you know, constrained

environment in San Francisco now. So that seems to be in your world, a developer doesn't have to review the code. And we seem to think maybe in the next under five years, which we call the midterm, I guess, in our industry, short term would be low number of years. Midterm would be five and long term be 10 to 20 decades. We're kind of in that midterm area, I think, based on what I'm hearing from you of I talk to it and it gives me the code back. That being said, the pace of code being written, just even doubling or, you know, if it doubles, we have like some kind of, if we just take Moore's law, every 18 months a developer's productivity doubles, that means we're doubling the number of programmers on the planet. That's pretty amazing. Does this mean the developer shortage will end? Does it mean developers will be less expensive and salaries will go down? What are your predictions, you know, as we move out of this towards, what you think is going to happen on a societal basis and our industry for that matter.

Yeah. I find it most useful to think of it as two points rather than a line and then draw the line after. I think right now we live in a world where humanity is constrained by human ingenuity and productivity where you need a smart person to come up with something and then you need a lot of smart people to come up with the details and build them and make them for an extended period of time. in hindsight, when you look at these things, you could always imagine like, okay, you could build this so much faster if you had already known all the things

that you figured out along the way and if your body moved a thousand times faster. So it is not inconceivable that you could build the things that today cost hundreds of millions of dollars to build for almost nothing if you take out those components.

And in a world where you have very affordable AGI. First of all, you have to make sure that humanity doesn't go extinct, because then nothing happens, and that's unfortunately a real risk. But if you get to the path where humanity is fine, I think it's almost guaranteed that we will end up in a world with virtually infinite ingenuity and productivity across the board. And it's very hard to imagine how fast such a world could move. even with things like Worse Law and the back of my head,

you know, you might imagine Worse Law every two hours. It's just those things, if you think from very kind of foundational principles, it is not inconceivable that if you had all the ideas that it took to make, you know, Apple's microprose, Apple's, I want shit, and you didn't know exactly what to do it, you could just throw the whole thing super fast, and you could run like a million copies of yourself that, like, can communicate at the speed of computers,

not the speed of human brains. There are many orders of magnitude in between at the throughput and the volume. The upper cap of this is unimaginably crazy. So I don't. And so now like going to the line, I think it starts with AI assistance. It starts with augmentation. It starts with things like you're describing as sort of a Moore's law of

in, you know, right now I would say the multipliers, even though things are flashy, it's quite small. I do think chat chabit did a great job at getting a lot of people's, helping people when they're stuck, which usually takes all of time, right, because it can answer questions about things that you might be stuck with.

Ultimately, we'll go from that to, okay, like, you just do it for me, why am I even doing this? And then once you get to the stage where you let it plan what to do and you don't just,

you know, you no longer just delegate microtasks, but you let it decide what they'll, then I just, this goes beyond anywhere, though. that we could draw up here. So developer productivity,

you know, I think you can draw these graphs for the next short while, but the world is just going to be an inherent, like a completely different place. We're going to have much, much more stuff, hopefully, much more sustainably,

hopefully distributed in a way that, you know, there will be still be some people who have more than others, but everyone will have much more than the average right now. And I think that the utopia version of this is amazing. And just almost going to feel like a different civilization to what we're used to.

I don't think there's a, I have so far always failed to draw a line. I can draw like a destroyed line and then like a big job. Yeah, I mean, our brains have a hard time. It took us thousands of years as a species to accept the concept of evolution.

Just that. That human beings emerged slowly over time from primates and, you know, mammals before that and, you know, et cetera, et cetera. Now we're on a timeline that's moving so fast, the concept of Moore's law moving from 18 months to 18 weeks to 18 days to 18 hours to 18 minutes to 18 seconds. It is conceivable, like you said, whoever the team was that designed the M1 chip

and the M2 chip, all of that knowledge going into a learning model and some artificial General Intelligence, it could make, instead of a new M2 chip every 18 months, M3 coming, you know, 18 months from now, it could make it 18 days or 18 weeks from now, and then we could have an M3 and M4 coming every day. And then we don't even know what a world with that much processing power would be like or that much software development would be like. We can't comprehend what it will learn about things in the real world. Medicine, fusion, energy, transportation, logistics, you know, it's unimaginable how quickly this could solve certain problems, longevity, diseases, cancers, DNA. We don't know what this speed level will do to these problem sets, correct? That's your interpretation of this. That's right. I mean, I try to think of what the limit is, and you can keep going, but I'll take it one step farther. Humans are currently made by humans

through a nine-month pregnancy process. There's clearly a way to piece together a few molecules out of pretty much nowhere through a biological process to create something as complex and amazing as a human being. If you look at factories that humanity has built for things like computers, which are probably the most complex piece of engineering

we've ever made, or for simple things, like I don't have a glass in my hand now, right? This is probably easy to make compared to a computer. They're just like nothing compared to what the human body can make. And that's something that was created through evolution, not through design. Now, imagine what you can create through design. If you had factories that didn't operate at the level of computer complexity, but at the level of human body complexity or beyond.

But now, instead of it being designed through an evolutionary process that has all of the flaws you could imagine, it's just pieced together of raw, out of raw atoms and molecules, whatever it is that you want to make. The universe is made of atoms, not of things, and we just see things because that's how we evolved. But the max level of craziness that we could get to

with this technology is, I don't know what you can build once you have machines that make things of the company. It's scary, it's really scary. But it's also, it seems like, you know, this is probably the most important thing for a civilization in the existence of that civilization to build, to build a thing that surpasses it. Like, evolution does that.

Evolution creates things that are more fit and by evolution's fitness function than the previous thing that was there, just through survival of the fittest. But never, ever, ever in history, has one of those things made a thing that is more fit? Like, never. But we're about to do that. And that loop is probably never going to stop, as you've alluded to, right? Like, if you make an M1, you get more computer, make an M2, that thing is real.

Like, if you have an AI that's a better ML engineer than I am, like, why would it be better at writing the next version, right? That's real. And, you know, again, that can go horribly wrong. If it goes right, it's just, it's a different civil. It's an evolutionary time scale. It's an evolutionary scale event. This is, it's not just the technology.

It's not like mobile or the internet. It's, those things are really important at the human timescales. But this is, this is like, there's a you know it's really interesting how you have as somebody on the inside understands this and and this is i think the problem uh we're seeing in the world uh you know when the ancient romans started writing books by hand and then giving them to the nobility right those books took months and months to be handwritten by monks or scribes whoever wrote them now if you were to take that book and transport the monk from 400 AD and put them into a modern printing press today

take their book, press a button and show them 10,000 copies of their book the next day. They would not know, they would think you were a god. They would think you were an alien. They would literally think you were Zeus and that you took

you know this book that had been written and took years or months and made 10,000 copies of it, they would not understand. It would be like Jesus taking a fish and turning it into two, but instead of two, it turned into 2,000, right? It would be like taking a farmer who was, you know, making, you know, pulling potatoes from a field and some Irish farmer, you know, pulling potatoes from a field and then showing them a million potatoes going through a factory onto trucks. They would say, how many humans do you have inside that box to do that? It would be hard

to comprehend. And this could be even more than that analogy. And as humans, we use analogies to try to understand stuff. That's, that's the analogy I've come to. It'd be just hard for somebody to understand modern day book printing or modern day newspaper printed or modern day food processing if you were but a farmer or somebody, you know, handwriting manuscripts. Developer talent is the most precious resource for B2B startups. You know that. And you want your developers focused on product, not on compliance, right? You want product.

You want to ship features for your users. And when you're selling B2B software to large enterprises, well, you're going to need to jump through a ton of security and compliance hoops. You know that. One of those hoops, especially for large companies, is the customers want to host your software on their cloud, and you need to build that out on a per customer basis. It's a lot of work, right? B2B startups constantly face this dilemma. Do you keep your develops focused on the infrastructure, which is going to hurt your product velocity, or do you keep them focused on

product velocity, which is going to delay your ability to close large customers? Well, there is now a solution for you. It's called release delivery. They automate the creation of enterprise class app delivery for private clouds and single tenant applications. This led you deliver your software seamlessly to any customer environment. And this is going to unlock a ton of revenue potential for you. Basically, all the big fish requirement of this. You know that. Release delivery is going to put all the tedious stuff on autopilot for you. You can turn your ideas into apps and deploy those apps quickly and flexibly. Let release show you the power of release delivery and get your first month free at release.com slash twist. That's right. They got that domainium. Release.com. What a great

domain. That's up to $10,000 in value at release.com slash twist. I like the analogy. I think maybe one way to look at it is that you move the person, like you move somewhere from the Stone Age to the point in time where he could write books. And it's like, okay, what is writing even? And then like 14 seconds later you moved them to the time where you could print 10,000 copies. And then 14 seconds later you moved them to the time where there is the internet, right? But you just keep doing that forever and you can't really ever grasp what's going on.

That it just keeps getting faster and crazier. We are already at the point. You know, things changing used to take. many generations, thousands of thousands of years ago. Innovation was so rare on the time axis that it was normal for a generation, like that many thousands of years ago, right? It was normal for a generation to just not have to change the paradigm. And even until recently, you know, the invention of things like the industrial revolution,

that took forever. If you look at the internet, that did not take forever. No, I mean, we take it, that did not take forever. Take a Gen Xer, right? Yeah. And compare what they experienced to what their great, great grandparents did in the 1800s, right? You know, they might have experienced the printing press. Hmm.

Nuclear power, maybe if they made it to the 50s. The machine gun. Trains. Maybe those were the biggest things they ever saw. Oh, and the airplane, right? Yeah. Pretty amazing stack for a hundred-year-old to live from 1850 to 1950 to have seen. They got to witness an airplane, a car, nuclear bombs, and the machine gun.

Now, what is Gen X experienced? The internet, DNA sequencing, GPS, you know, I mean, and now AI, right? Like, it's a whole different cohort of... And it's not over. Well, and it seems like it's going faster. Like, what I just described would be, like, one item, right? Yeah. Yeah.

Exactly. Oh, and self-driving potentially and life extension. You know, like these things, it's just unimaginable what this could do

in the midterm to long term. When you saw, I don't know if you saw yesterday, IBM's like, you know what? We're pausing, hiring about 30% of new jobs, thousands of jobs,

because we actually think they're going to be these back office roles are just going to be all AI. so what's the point of even hiring them now? We'll just pause hiring them, knowing that AI is coming so quickly. Now, this is IBM that created Watson,

deep blue, you know, B. Casparov, like they know what they're talking about. You know, you can make fun of IBM if you want to, but they actually have been a pretty storied institution. Yeah. And for them to just say,

what's the point of hiring operations people? It's a pretty big statement. You know, I don't know if they're blaming the AI. Maybe they're blaming the AI. the AI. I don't know about the specific story, but I would say that generally, you know, people say a lot of things, PR, whatnot, but in those actions, you really see what people

believe. You know, a public company allocating funds one way or another is usually associated with their beliefs of where the world is going. And yesterday, you know, there was an earnings report of a company that lost revenue because of Jash, VPT and that these things are just going to increase. massively. is the company you're talking about. And Chag is a student resource for getting homework help to solve problems,

writing, citations, you can rent textbooks, all of that, those study guides, exams, they said people are just using chat GPT now. So students are going to find the most affordable, free, in fact, solution to their problems. That's why they will photocopy a book, you know, or a chapter of a book or share a book or buy use book. So they're just going to go down, like what are going down to hill. It's going to find the most optimal route. Yeah, Chag was the company you're talking about. Yeah. It's crazy. I forgot the name for a second. Yeah. Thanks. Did you seem to have a positive view of all this? Well, look, I studied game theory

for many years. This is the field I did research in since I was 15 as just the thing I kind of, I don't know, I like decisions. Humanity is just a large collection of decisions and the incentives and rules around those decisions. And a lot of, you know, kind of group behavior is predictable by those things. And a lot of limitations on group behavior are predictable by those things. So I'm a fan of that. Am I optimistic? Well, let me be clear.

I think there is a good path. And that path is mind-blowingly amazing. It's an evolutionary timescale thing that we are just immeasurably lucky to be experiencing if it happens. We are like, maybe the generation people are going to envy forever. You know, the last one to experience

non-AI-life, the first one to experience AI life and the whole transition. I mean, this is probably really the most exciting part of history to ever live in, and maybe forever, who knows, just in terms of how exciting it is. But at the same time,

you know, humans and ants, you leave them alone when you're hiking in the forest, and you're kind of like, you're cute, but you don't really care. And when they're in your kitchen, they're just dead, like a half a day later, because you have some ant poison, and then they don't understand what you're doing. They don't understand how ant poison works. So they just, you know, they go there, the way ant poison works is they have to put a little thing

somewhere, and then they go there, they carry that ant poison back and all the ants die. They didn't even know what happened. It just, they're not dead. That's what happens if a dumb species annoys a smart species. and if we build AI that is so as much smarter, then we are as we are smarter compared to ants, and it doesn't like us, or someone tells it to kill us all,

and there's no way to defend against that, which is both a game theory problem and a simple exor, is it even possible, like, similar to cyber defense, where it's just not clear how hard attacking versus defenses in some vectors. That is real. And if it's extremely high variance in that I believe there are ways to do this that lead to the good future and there are ways to do this that lead to everyone die, I do, I think both are possible.

I have a lot of thoughts on the split in the tree at that junk shark there, but I don't know if I'd call it optimistic. I'd say this is the most important time to ever live and I don't know why, but glad I have some ability to contribute to hopefully making it go well. Let's hope it's not a zero-sum game. If this is a

like a giant 10,000 piece puzzle, hopefully the AI and humans are solving this puzzle together and that whoever puts the last piece in, we both benefit from the puzzle being completely. The puzzle being humanity understanding the nature of consciousness, the universe, the cosmos,

why we're here, what all this is, and then having some joy from the experience. And, you know, I don't actually define it as living forever, but hitting 52. I'd certainly like to double my lifespan and maybe preserve consciousness and understand what the hell is. It does feel like this is going to tell us. But I love your analogy of, hey, let's be thoughtful because this thing could grow so, you know, exponentially. A 16th and 18 second, Moore's Law is like lighting a fuse to something that we don't know the chemical nature of. And it's literally like you could be lighting a fuse to a giant bucket of kerosene underneath you and not know it.

and by the time you figure out what happened, you're incinerated, right? Like, that is kind of the analogy here. And I think we have to be thoughtful. And the fact that you're recognizing that, I think, are you for slowing this down? Is there any guardrail here? Should people who work on this technology be licensed like doctors are or truck drivers

are? Should people be required if they're using these tools to have, you know, some ownership of what they're building? You have any thoughts on when regulation should come to this space? If at all? Yeah, I think regulation is needed now. I think that would be a good thing.

The making it, I do think you need to do this work because it's ultimately, I do believe that if something is possible, it will happen. You can't live in the world where all of humanity knows how to build a screwdriver and nobody builds a screwdriver. It's just not going to work, right? someone's going to hide in a basement and build it. And with compute getting cheaper, training a GPD4 quality model will get cheaper. And, you know, it will eventually happen. So we need to make sure that it gets developed safely. So both that it a gets developed and B safely in the same sense, you need to develop it

to develop it to develop it safely. Otherwise, I just can't see a world where we stop it or pause it. And again, there is a utopia version that I would certainly love to be in. But yes, I do believe you should make it extremely, hard to mess up. Life is fascinating. We should just not much enough. What are the choke points here?

I mean, one I can think of is access to compute power required. People, and then maybe the people who do control that compute power, I know it sounds silly, AWS, Nvidia, Azure,

they control the means of deployment. That seems to be a place where you could be thoughtful, just like people who control the precursors to certain chemicals that are dangerous, have some controls. And for people who are in the no control camp, go try to buy a bunch of fertilizer after the Oklahoma City bombing. If you want to go buy a bunch of fertilizer, it's tagged.

It has secret codes inside the fertilizer. They report back to the government when somebody comes in and buys a lot of fertilizer who's not a farmer. There are some controls in place. because people realize fertilizer bombs can knock a building down. It's not controversial to think about regulating

a certain size of a gun, a certain caliber of gun. You can't buy a 50 caliber, I don't believe. You can't buy 50 caliber, certainly machine guns, automatic machine guns. That's limited to the military.

But you can buy, you know, 38s and 45s. So there is a certain caliber under which, you know, some regulations occur. How do you think about specific regulations? I put one out there, which is the deployment, and access to deployment technology. If you...

I'll be careful. So first, I think training is something that is a huge showpoint because it right now requires a whole lot of compute that you can't hide to really, to train these models you need very specific hardware and only a few companies are building. So if you have those companies follows very strict procedures and who gets access and, you know, what monitoring should look like. I think disclosure is important to just get an idea of what's even going on. Training is one that's easier to regulate.

This starts to change once, you know, what happened with this model called Lama, or the way it's just, you know, where like kind of half open source, then they leaked and everyone has them. Once that happened, I think you're in big trouble because inference is not that different. difficult. Even if you, like, so, you know, if I downloaded GPD4 on my computer, like, I could figure out how to serve it without anyone noticing regardless of government constraints. It might run extremely slowly, frankly, who cares? So, oh my God, my completion takes 15 minutes instead of three seconds. Like, what's 15 minutes? Nothing, right? So I can run that. I could run that on my CPU with hard drive offloading, which is, for fancy terms, but the point is like, you can run their stuff on any hardware, it's just going to be extremely slow. So that I don't think. I don't

think you can particularly stop individual deployment. You can stop rollout, you can stop productization, which I think a lot of the LLM companies are doing themselves, which is good, from making sure that, you know, malicious use cases are not being pursued. They do that well. But training is the big one. And I think for that, that's why I think regulation should come soon. For that, for that, to work, it just needs to be done before there are a lot of models like that. And you need to make sure that, as you said, I think licensing of companies and individuals that are in control of that and then the security procedures to ensure that the weights do not, in fact, get distributed to the world are important. I mean, you know, you'd bump into many international issues there, like,

what do you do if other countries don't want to comply with that? Like, do you just slow down as the US? I mean, I'd understand if the US doesn't want to slow down if other countries don't. So this needs to be a multinational thing, just like with nukes. And there's a long story about how it went with nukes. I don't know how much time we have. But basically the whole, like, we don't use nukes thing fell through because they couldn't detect, I think it was underground tests back then because they didn't have a measurement capabilities that were good enough.

So if you can't detect it, you know, people don't trust each other, governments don't trust each other, and then you get everyone doing it anyway. But, you know, training, I think, you know, I just because they're very few providers. I mean, yeah. Yeah. We're supposed to not be doing bio-weapons research, right? Like, I think there's some international treaties on this, and it's like, is Saddam Hussein or, you know, Kim Jong-un or, you know, some bad actor? Are they going to follow it?

No. And can you detect it? No. Can you maybe get spies to tell you what's going on? Perhaps. Can you find some precursors and follow them around or do some lab detection, I guess? but yeah, brave new world. But people should just take note.

I mean, the fact that, you know, people who are building this stuff, Jeffrey Hinton, the guy who's just a godfather of AI, of course, but he drove this, that he's worrying of it. And everyone, I mean, I'm just a tiny fish in the pond. The people who've been working on this for decades are worried. This is different from when people are worried about, you know,

crypto, like, or when people are worried about whatever else is the thing they're currently worried about, right? that the people who have been working on us for ages who have dedicated their careers to it, including myself, are saying that this is very much a path with a juncture. And we are supposed to be careful here.

And, you know, we have government authorities checking on development of other potentially dangerous technologies and tools. And we should just have that here. Yeah, if you don't know about Hinton, he has left Google with a warning of danger ahead. And this is a reoccurring theme, people who helped build this saying, this is moving at a pace that's making me uncomfortable. Thank you so much for sharing what you're doing.

I wish you great success. And we'll have you back on the program after you'll long. launch and you get a, you know, I don't know, a couple more Moore's law cycles. Learning's. So I'll probably have you back on the show in 18 minutes to explain how this all turned out for humanity. Great having of the show. I know you raised a bucket load of capital.

Are you hiring? And yeah. Thank you so much for. To join the team. Yeah, this is a great conversation. I'll be totally honest. I was like, wow. Okay, this is really simple.

I get it. a totally logical idea for a company. It's going to be very successful, to do a great return for investors. And then we started talking about the big picture, and you turned out to be a wonderfully thoughtful guest. And we had some really great moments here of like thoughtfulness, just batting back and forth,

you know, some analogies of where this is going. And you really expanded my thinking on it. And I'm sure the audience will get some great usage out of your thoughts and your thoughtfulness. Where can people find out more about your company, sign up for the beta, maybe join your team.

I just really appreciate what you just said. Thank you. I'm glad I can contribute some amount to hopefully, you know, the wider picture of VGI going well beyond me just pressing buttons on my computer, which I certainly like doing and do a lot of. But, you know, okay, so we're, yes, we can find us at magic.dev,

DEV, like, you know, developer. And joining us, yes, we're looking for people who care a lot about AI going super well and not going super bad. and who are good at pressing buttons that make computers think, you know, the deal of you, if you are one of those people. So just reach out at Eric at magic.dev,

and I would love to hear from you. Magic.gov slash join as well to see what they're looking for. And you can just email. And, yeah, no, I'm an open inbox guy. Love it. Slash waitlist if you want to use the product. Maybe, I don't know when this is, so maybe Windows comes out, potentially is already used,

but probably waitlist. We'll see. All right, everybody. That's Eric, magic.dev, and we'll see you all next time.