Introduction

So my name is Matthew, and I'm an entrepreneur who started a couple growth marketing agencies and a couple software companies.

I spent about seven years in the blockchain space. And more recently, I've been getting interested in AI, of course. starting to try to figure out sort of where this is all going.

So as someone who's sort of made a career out of trying to work on the next thing, I'm at a stage where I'm trying to think about what that could be. And so that's what I've been trying to think about, especially where I could maybe like add value.

Sharing Research and Insights

Like I said, last month was my first time at this meetup, and then I kind of shared some of the research that I was doing with Josh and team, and he was gracious enough to invite me to share it with you. So I'm still working on sort of what to call some of this stuff. You know the quote from Mark Twain, like, if I had more time, this would have been a shorter deck. But that's what you should strap in for.

I just have a bunch of slides. It's going to be like screenshots of companies' websites and me telling you what they do. But it's all going to be, I think, connected to some themes that I think are interesting.

I could have called this the age of autonomy or the age of acceleration. I don't know. But anyway, I also have these fun mid-journey images that I made along the way. And they're just whatever showed up. So don't judge them.

It's just me trying to add some humor as I move along. So I could have called this drones and clones. I was also thinking about, you know, very futuristic things. So we'll come to that.

Science Fact vs. Science Fiction

And I think the first theme I wanted to kind of talk about is science fact and science fiction. So some interesting things I think people are kind of talking about in the zeitgeist. There's this one quote that I think is not... Unique necessarily, but it kind of hits the nail on the head a little bit for me.

So American science fiction in the 1950s was the most inspiring thing in the world. What happened? And there's actually some studies I've seen that have tried to like analyze science fiction over the years and found that not only has it turned more dystopian, but we're actually just publishing less of it in general. As if like we sort of don't know what to write about.

We can't seem to predict very far into the future. This is Beth Jezos. If you've maybe heard of him, the founder of the EAC movement, Effective Accelerationism. I recommend checking them out. They're pretty, I think, kind of rad.

So along these lines, you may have heard of Ray Kurzweil's book, The Singularity is Near. And I just want to talk about maybe we're already on the cusp of it. It turns out that he actually predicted human-level AI would come in the 2020s back in 1999. So here we are. Certainly, he seems to be on track.

He also said that by 2045, which is what he was calling the real singularity, we'll have increased what we know as humans by a million times. And also every computer will know everything that every human knows. So maybe the point of all this is it's hard to imagine what the future is going to be like. And that's why we're seeing kind of this dearth of sci-fi.

It's also interesting to know, you probably heard the phrase, the singularity is near, but the subtitle to that book is When Humans Transcend Biology, which is going to kind of fit into this talk a little bit too. There's some books I'm going to recommend along the way if you're into this stuff. One of them is called AI 2041. So it's actually a really interesting idea. So it's 10 short stories.

One of them is written by two people. So one is the former president of Google China. and also the co-chair of the AI Council at the World Economic Forum. So he's sort of the brains, if you will, who actually knows sort of where AI is going. And then he was partnered with a creative writer, the president of the World Chinese Science Fiction Association, and they wrote 10 short stories predicting what the world would look like in 20 years. And starts off with sort of a very analytical description of some topics he wants to introduce, a short story that kind of gets the idea across in a compelling way, and then some analysis at the end.

So I recommend you check this out. But again, this theme is sort of science fact and science fiction. Another book I recommend is called Life 3.0. So it's written by this MIT physicist and computer scientist. It was written in, I think, 2017.

And when you read it, it feels like it was written yesterday. And it actually starts with this really fantastic short story about this company called the Omega Team that kind of gets just a little bit farther ahead than everybody else in AI and then ends up kind of taking over the world. And it's a compelling possible vision.

But one of the things, again, that's going to kind of be a thread through this talk is he talks about life 3.0. So basically, life 1.0 is, you know, primordial bacteria that sort of its only way of improving is through evolution. So every generation dies, right? And then evolution is picking the natural selection is improving it. So it has to go through generational experiences in order to improve.

Life 2.0 is where we can improve ourselves while we're living through our intelligence. And so we can learn from past humans and we can adapt much more effectively than we could in the past. And so his word for that is sort of upgrading our software while we're still alive. Life 3.0 is when we can upgrade our hardware. And again, to be talked about.

This former Googler in the AI space, Ramesh Naam, wrote a sci-fi trilogy. The first is called Nexus. Highly recommend it. Again, very prescient. I think it just had its 10 year anniversary.

It's about a drug that's invented, a drug that's actually like nanobots that you swallow it and then they implant in your spinal cord and then suddenly you have the ability to sort of talk to anyone else who's taken the drug through basically internet connected telepathy. Very cool. He also has written basically fictional, or sorry, like essays, nonfiction, one of which is called More Than Human, and he explores this notion of transhumanism, basically what would it be like if we used technology to really seriously advance what we're capable of. So I'll come back to this.

Another book is called The Coming Wave. So this is by the co-founder of DeepMind. He posits that there's two huge waves that are coming. One is AI, which of course we're talking a lot about. The other is synthetic biology, which I'm not an expert, but basically I understand it as the ability to generate

you know, cells or organisms that themselves can produce things we want, like as a byproduct. So imagine, for example, an organism that eats carbon out of the air and makes water. I just made that up. But that's sort of the type of thing that synthetic biology is looking for. You would give your genome into a machine, and it could make you a liver if you needed a liver transplant. It would be your liver, not having to wait for someone else. Synthetic biology.

The other complementary waves, this is all predicted to come in the next couple of decades. The other major waves he talks about are energy. Major advances in this would be fusion, fission, batteries. Maybe you saw the awesome story about we just discovered the largest lithium deposit in the world in California, which is a very important resource for batteries and electricity and solar itself. Robots are coming online which is a common or a sort of fun phrase for this is embodied AI and then quantum is like taking computing to the limits of physics itself so

AI and Deep Tech

to the more grounded theme that I wanna talk about is AI and deep tech. So deep tech is this kind of emergent buzzword that might frankly be done three months from now, I don't know, but at the moment it seems to be gaining some traction and I'm finding it fascinating and I'm seeing a lot of complements or sort of areas of opportunity in this intersection.

And so to define it, so I'm going to take you through some categories. But broadly speaking, they are projects based on real breakthroughs in science and technology. They don't have to be AI at all, as there's some examples of that that kind of snuck its way into the deck, but that's where I'm going to try to focus is this intersection of AI and these areas because this is where the greatest acceleration is happening and is expected to take place.

So why is this the case? So you've heard we all have some grounding in AI, so I'm just going to do a high level on this. And this is just me positing. One is unique data sets. So all of these AI tools are really, if you're thinking about AI companies trying to start something and it's not just going to get devoured by open AI tomorrow, one of the things to keep yourself competitive is to go after unique data sets. And these unique data sets, are prevalent if you're working in cutting-edge science and technology. You're working with genomes, you're working with material science, you're working with a variety of things that have nothing to do with, like, text that's on the internet or YouTube comments. And so that in and of itself maybe gives you, you know, white space.

Generative AI. So you can not only train models on making poems or ad copy, but also what you ask of it is, can you give me a molecule that can withstand the temperatures of nuclear fusion so I can line my reactors? And it generates a bunch of possible 3D molecules that can do that. And then you go test them in your lab.

I'll talk about more of that in a second. And then lastly, autonomy. So deep tech processes are either extremely difficult for humans to do or impossible. or they need to be automated at scale to make them capital efficient.

Categories in Deep Tech

So either way, there's a huge sort of demand from deep tech efforts to use AI to automate. So here are the six, and I have a seventh that I'll drop in at the end. But here are the six categories that I'm tracking to basically build, I have like a watch list, and I'm gonna show you some of them, of companies that are sort of operating in this area.

And again, it's just inspiring me, so I'm here to share them with you to think about and to discuss. So of course there's computing. I'm going to kind of go quick over this. So we know there's obviously models and the models are changing all the time. This quote from Elon just recently compute coming online is increasing 10 X every six months. That would be 5,000 X more than Morris law, which is doubling every two years. And from the same Karpathy LLM that we heard about, like intro to LLM video we heard about earlier, There appears to be this linear relationship. Basically, the more data that you can just throw at these things, they don't seem to get worse in terms of a diminishing return. So there's a huge belief that there's an ROI if you just keep throwing more data at these things, which is why I see Sam Altman raising maybe $7 trillion.

So anyway, that's models. Then there's chips. You've maybe seen Grok. You can go to grokwithaq.com and try this stuff. It's pretty cool. When you put in a prompt, the thing will return a response. in a fraction of a second, whereas you're used to ChatGPT taking longer. It really doesn't add that much value today. But it just shows you that it's because they've created a chip specifically for this. And so we're just getting started with this stuff. Xtropic is a company that raised $14 million. This is the guy behind EAC. And he's talking about creating a basically quantum-ish type computer so that it can optimize for this stuff. So there's just going to be more of this.

Quantum and Material Science

Quantum, I didn't think this stuff was ever going to be in our lifetime, but it apparently is. There's a company called Quantinium. They just raised $300 million at a $5 billion pre-money valuation. These things are shipping now. You can buy these quantum computers, which again are supposed to be sort of many many orders of magnitude better than what we're dealing with today. So there's plenty of stuff going on there and that brings us to I think the next kind of core tangential category which is material science.

So people are trying to figure out, there's a lot of text here sorry, how to make better materials for the valuable future things we need to make, whether that's chips, whether that's robots, whether that's nuclear reactors, whether that's rockets, all these things. And so some of the interesting things people are doing is actually using models to generate these ideas. where they're actually generating, almost subverting the way we think about the scientific method. Typically, people would have an idea for a material that might withstand or would have the properties that they're looking for. You'd go into a lab. You would test it for those properties. You'd be right or wrong. And you'd have to try on your way through that, like Edison. This is basically saying, why don't we run as many simulations in a VR type kind of situation as we possibly can off of ideas generated from models trained on these specific scientific characteristics and only then take the ranked best options that the AI gives us into the lab to test those. And this is already being done and shown to make massive acceleration in terms of not only how many things we're discovering faster, but at a cost that is much, much cheaper than it was before. So where there's strong areas of interest, silicon and computer chips, carbon, carbon capture, metals for electricity, construction, robotics, et cetera, et cetera, et cetera.

Robotics

Robots. Figure just announced a massive deal with OpenAI, Tesla. Don't forget about Boston Dynamics. You know that spot dog?

So what that thing is hired for now is to basically walk around factories and industrial parks and make sure that nothing's broken, and alert people if it is, and to go climb up really dangerous areas and check on those things. Great. Already in use. Already deployed.

You've probably seen this Agility Robotics. This is basically Amazon warehouse workers. This is their biggest investment in robotics.

What this screenshot shows is that this robot was just placed in a room with these three colored boxes on the back row and then four columns in the front row and given an LLM direction. Take the box that's the color of Darth Vader's lightsaber and move it to the tallest tower in the front row. And it was able to figure out that that meant red, go pick up the box, go put it on the tallest column in the front row without any pre-training.

This is a warehouse, they just completely transform how you track all of your packages. This is, imagine the task of you have a microscope and someone's actually bent over a microscope trying to see if this petri dish has what they're looking for. This, you just slide it into a thing and it does all the computer vision for you.

uh... this is a little robot that goes around retail stores and sees if items are like you know when an item is like out or if it's like maybe all the way in the back and someone's always kind of walking around like pulling it to the front apparently that actually has a big impact on whether the item keeps selling so this robot will go around and alert the staff so the staff is not randomly walking around to replace stuff but is getting basically like uber driver dispatch to go refill things this way. It also allows for a customer to come up and ask questions with an LLM response.

There's drones in the ocean that are using solar power to just always be on and be used to map the ocean floor and various things. This is a submarine that does the same thing. There's planes that purport to basically soon be able to take you from New York to Paris and they don't even have a pilot, it's a drone. These are funded companies working on this today.

Space and Defense

That's robots and drones. That takes us now to a few more specialized areas I want to talk about.

Space and defense. Neurospace.

There's a huge number of companies that are basically helping you manage your satellite fleet. In order to make sure that either you constantly have coverage or at the very least you're never running into other people's satellites. So they're using AI for that.

This is Relativity Space, which is a 3D printed only rocket company. So just bringing back the material science angle. So again, they're trying to figure out how to get the cost of rocketry down even further and using material science and 3D printing to drive that.

Varda is a company that launches satellites to space, uses robots to run basically lab experiments in space that would otherwise fail on Earth due to gravity, and then bring the stuff back to Earth to be used. This is kind of creepy.

So this company, basically you can take existing F-16s, etc., put this technology onto it, and suddenly it's an autonomously flown fighter jet. And you can actually have them basically operate as swarms. And their thesis is that this is the way that air warfare will happen in the future.

This company, similar idea, so this idea of deterrence or security. So this is basically a drone system that your public police department can use. And anytime they get a particular 911 call that may require certain elevated response, it'll dispatch a drone fleet to that area so that it has constant coverage of basically video of that space to support the team on the ground, all autonomous.

Similar idea, this is everyone a fan of Minority Report would be happy to see the words proactive security, a little scary. But basically this idea, same thing, that if it detects certain suspicious things it will just automatically alert the police. Same thing here.

This thing on the right is a solar powered face detecting camera. that is turned on and license plate camera that's turned on if it detects a gunshot or other sort of emergency type sounds and then provides information to the police and surrounding communities.

The AI assistant for every 911 call. You kind of get the point here.

Agriculture, Food, and Climate

Next category, ag, food, and climate. So this is a robot that helps farmers. This is a collar that cows wear. And you can see this dotted sort of fence.

What it's doing is actually using GPS. And I believe it's imaging how much of the grass the cows have eaten. And then it, through sound, moves. It's like a digital fence, basically.

It moves the cows to other parts of the pasture through this collar to make sure that they're basically both secure and rotating where they're eating. This is super cool company.

So apparently one out of three fishes that's harvested, only one out of three fishes that's harvested makes it to a human's plate. A vast majority of the waste goes to the fact that most fishes are killed due to suffocation or like a toxic injection on the boats. and either way that has like a terrible end of life sort of experience for the fish that puts this kind of acidity into the meat and then it like isn't good enough for humans so it just doesn't get sold.

This company is basically using a robotic fish butcher to precisely butcher the fish the way that a sushi chef would, and therefore increasing the yield, of course decreasing the waste, improving the humaneness of this. Again, you get the idea, pretty cool.

Ohalo is... using some of the generative AI elements I told you about to explore how to create new varieties of crops that are either better yielding or better nutrients.

This is a leather company that is basically making synthetic leather, but it's actually protein. So it's actually a biomaterial. Again, they're using AI to help them discover how to produce these things, but essentially it allows them to create leather quality material that is not actually using the animals and so on.

This is one of my favorite companies. This is a recycling company where they're using, you put a huge bucket of recycling, indiscriminate recycling on a conveyor belt and then it uses computer vision to detect what the different materials are and then it's rigged to like a little air hockey almost table on the conveyor belt that with little air jets pushes the things around until it sorts everything. into, as you can kind of see with the picture behind them, you just get these cubes of sorted recycling on the other side. And there's no human intervention in between.

It's just a fully automated recycling sorter.

This thing, these guys just took an excavator, like normal off-the-shelf excavator, and put, again, computer vision on it with a pile driver on the front. And they are tackling this issue with building out solar farms where there's just this repetitive task where you have to just pile drive these steel rods into the ground to put the arrays on top of. And so a whole human crew can get 100 of these things done in a day. They have one rigged excavator that can basically use computer vision and GPS to know exactly where it is and do the work. And one of these things can do 300 steel rods in a day. And so they're viewing it as an improvement, obviously, on the build-out of solar.

This company wants to make rain. I'm just going to keep going faster. Okay, I'm almost at the end.

Life Sciences

Life science. So... Same idea of the material science sort of gen AI. There's a lot of people trying to create basically drugs, like accelerating drug discovery.

And so this company has received, I think, over a billion dollars in funding. And this pitch is AI-enabled, as you can see, precision medicine. Their key focus is cancer.

Same. So huge. There's so many companies that are essentially doing the same thing, which is using AI to accelerate drug discovery, especially with this kind of related idea, gene and cell therapy, which is personalized medicine, the idea that you would actually be able to encode your entire or get your entire genome read. And then the medicine that's made for you is maybe like personalized medicine, a generative model for protein design.

So I'm kind of showing these things just to show how big this space is. Cool.

Then there's neuromodulation. So I'm starting to get to my last sort of category. Then there's people who are creating basically wearables right now that are using your brainwaves to modulate your behavior, your emotions, et cetera.

And so what they're doing, again, where the AI comes in here is that people have been able to isolate what brain waves from what areas of your brain are associated with what behaviors or, or sort of, you know, vision, et cetera, et cetera. So they're able to isolate like, okay, this brainwave activity right now is associated with Matt's, um, stress or something to that effect.

And so what this company will do is when I'm wearing this, or in this case it helps with sleep it says, and a few other things. When I'm wearing this, if you think about brainwaves and you know a little bit about wave science, if two waves hit each other at the peak of their wave, they double. So they have constructive interference is what it's called. If a wave at its peak hits another wave at its trough, they cancel. And it's called destructive interference. So there's suddenly no wave. So what they're doing is if you're trying to go to sleep and that feeling where you're closing your eyes and you can't quiet your brain, they're isolating that brain wave and then running a destructive interference wave back at it, essentially quieting your brain. And they purport to help you fall asleep in five minutes or less when you wear this. Hopefully you wake up.

Transhumanism and the Future

So that brings me to the last category, which is transhumanism, which is weird and out there. But I think we should all be prepared for it. Because I think this whole space, as I've taken a step back, the future is much closer than I originally thought.

There's been tons of investment in a lot of areas that have been quiet for a while. But now the AI, I think, is having its moment. You're noticing how fast they're accelerating. And they're all benefiting from this new AI. technologies.

Every one of these sectors is having accelerations because of some of the recent stuff that's been happening outside of all the consumer interest that it tends to get. So, leaving you with this.

Company called ORCID. So I am a parent of two. I don't know if any of you have parents or kids, rather. All of you must have parents. We're not there yet with transhumanism.

But if you've had kids, then you've maybe had this discussion with your practitioner where you may carry genetic diseases. You may carry genetic issues. And you have an opportunity to get those screened for. And then you have an opportunity to do something about it if that's what your family wants to do. That tends to be through termination.

Again, if you're going through sort of a live birth kind of situation, or sort of natural birth, I should say, a natural conception. There's a number of people who are infertile. or for a variety of reasons want to explore egg freezing or just delaying sort of when they're going to conceive, or they already know that they have one of these genetic diseases or that they are a carrier of it and maybe they're dormant and they're afraid to have a child as a result.

So those people, some of them, elect to have IVF, in feature of fertilization, which has been around for a long time, or long in terms of science, and at this point now five million Americans have been born through IVF. And about 2% of American kids today are born through this. And so when you're born this way, you have an opportunity to screen the embryos before reimplantation.

So again, in the traditional way of conceiving is you get pregnant. And then you can, at that point, screen the embryo that's already in the uterus. And then again, you have a choice to make.

In this case, because it's in vitro, you have typically about five or six embryos per cycle. And then you choose which ones to put in for re-implantation. This is how it works.

Previously, you could only screen about 1% to 2% of the genome for a good chunk of the diseases that we know of. This company, ORCID, due to AI advances, is now the first company that offers whole genome sequencing of embryos. So it can tell you everything about this embryo, basically. They also are able to basically capture as much of the DNA as they need from the sort of egg retrieval process in order to do this.

What it means is that they can now basically tell parents much more than they previously could. And so they have more confidence to move forward without maybe introducing any kind of genetic issues, which is amazing. As you maybe are already sort of thinking about where this could go, this is like very close to basically designer babies and the idea that you'd be able to actually see like a projection of what your kid could look like. You would have like these abilities to possibly even, you know, select for certain traits like intelligence or strength or things that sort of fall outside the category of like disease. But you would certainly have the information available to do that if you wanted with this kind of company.

So the future's coming. This is a section, I won't show this to you now, but this is basically written from that book I mentioned and says exactly what I just said, where you'd be able to see and you can, if you're scanning it, you can imagine this even with some of the gen AI ideas that you get generated pictures of what the kid could look like when they're teenagers, just based off of their DNA screen, pretty wild stuff.

Next, advanced prosthetics. If you've been paying attention, robots are pretty damn good. And when we get prosthetics, we've seen some of the ones that are obviously like the blades in terms of helping really go fast. But they're not really using technology. They're not really integrated. They're not necessarily IoT devices. They're showing that robotic. until maybe something like this.

So this is a company I came across called Atom Limbs, and I recommend you check this out. It's basically being able to put a robot arm on, and again, using AI, you're able to use your brain waves to basically think it and use it. Neurotechnology, which is just taking a step beyond neuromodulation, which is sort of a wearable to these companies that are basically potentially even getting to implantation.

Conclusion

I'm coming up on my time. So to end it, I think that there's huge amounts of opportunity. in all these spaces. And in particular, I encourage, you know, a city as diverse as New York with so many things going on, there's like the obvious sort of sectors to focus on. But if you're not familiar, there's even an organization called New Lab out of Brooklyn, which is focusing on the kind of like deep tech things. So anyway, I'm into it. If you're into it, get in touch with me. I'm going to continue to sort of share my research as I go. So follow me. And yeah, thanks for listening.

Thanks.