Pablos Holman on AI, the Future, and Why the World Needs More Energy

Reason: The book is subtitled Creating Technology That Matters. What is technology that doesn't matter?

Holman: We live in this world we think is full of technology, but it's mostly just full of software. I think we're setting our sights a little low the last couple of decades. If you just have iPhone apps to have weed delivered to your dorm room by a drone, that doesn't really feel like technology to me. Meanwhile, it's taken our attention away from other technologies that could make a bigger difference.

Nobody loves software more than me. My best friend from childhood is an Apple II. It's turned out to be this incredible tool that we can use that's generally applicable to everything, and that's important. But now that we've done software to everything, we need to get back to doing the other stuff.

So when Marc Andreessen says that software is eating the world…

I say the world can't eat software. If you think about food, clean water, sanitation, construction, manufacturing, energy, all these things every human on Earth relies on, you won't radically improve them with software. You can make them a few percent better, and we are doing that, but you're not going to make them 10 times better. That's what I think of when I'm talking about deep tech—technologies that could go after those bigger problems.

You run a fund called Deep Future. What are some of the companies you're investing in?

We invest in these mad scientists—they're coming out of a lab and into a startup, and most of them have some kind of breakthrough or we wouldn't bother. One in New York. These guys figured out that plants protect themselves from bugs using this mushroom spore in nature. They've been able to take that mushroom spore and evolve it to target different invasive species of bugs. They're using it right now to save bee colonies from the Varroa mite. The reason bee colonies are wiped out is this little bug called Varroa mite eating the bees. Well, they put a teabag of mushroom spores in the beehive, and it will wipe out the mite and save the bees. It's this amazing, beautiful thing, but it's not software.

What about the Roman concrete? I was in Rome earlier this year for the first time and was amazed looking at the Colosseum, which is still around after over 2,000 years.

Or the Pantheon, which is even more miraculous. If you've seen that thing, it's like the dome building in Rome, it's 2,000 years old, it's made of unreinforced concrete, and it's in a seismic zone. Everything we build out of cement—which is almost everything we build—is made of concrete with steel rebar to reinforce it, and then it crumbles in 50 years. Nobody's ever been able to figure out "How did the Romans do that?" I found this guy at MIT who figured it out. Now we can make cement that lasts virtually forever, use less of it, use less steel, and the kicker is, it's less CO2. There's nothing not to like about this. And you could do it at any cement plant. It doesn't cost more; you go in and change the formula a bit.

Now they're in production. They're building stuff with this cement. [In] less than 10 years, we can probably upgrade all the cement being produced.

A lot of the companies that you discuss in Deep Future have to do with energy. You talk about a company that is going to use solar power by being in space where the sun never sets. Can you explain what they're doing?

If you just look at what's happening with our solar farms, we keep making more and more of them but the relentless onslaught of night keeps fucking with our solar panels. But if you take the same solar farm and launch it into space, it'll get sun 24/7. It's actually noon in space all the time. If you put a solar panel in space, it'll get sun all year long—eight times as much energy—and then you can beam it down to Earth using radio waves that go right through clouds. This sounds like science fiction, I know, but it's real. All the technology, we have. They're aiming to put the first commercial array up in four years.

Is there a regulatory regime or exorbitant costs slowing it down or preventing it from happening?

We actually have a pretty functional regulatory situation for doing things in space. So that's less worrisome than it has been in other industries.

The cost of putting stuff in space was just laughably expensive for our whole lifetime. In a space shuttle, it would've cost $40,000 to get this book into space. Now it's about $1,500. But the target that SpaceX has for their big rocket is $10 a kilogram. In this lifetime, you'll store your old sportsball gear in space instead of your garage or closet.

Do you really believe that?

Yeah. It's called space for a reason. We're really going to do that.

You determined that energy production is the most important thing to improve life on Earth for the most people. Is that because as we get richer, we consume more electricity?

We consume more energy because we're rich, but a lot of people don't have enough. They need to consume more energy, not because they're rich, but because they need it. If you average global energy production, let's say you get about one toaster per person—you duct tape the button down on a toaster, run it 24/7—that's about how much energy the average earthling gets. Americans get eight bonus toasters. You're literally getting an insane amount of investment of energy in every single American to get those averages. Three billion people live on less than one toaster.

We've made enough energy for all the people in the West and north of the equator, but we just haven't finished the job. Less than one toaster is not an acceptable living standard. You want to get somebody up at least to—Europe used to be five, six toasters, now they're more like four or five, but that's an acceptable living standard. To get those averages up, we've got to 10X global energy production. And that's kind of heretical. Most people want to show a couple of percent a year of growth.

Or they want to manage demand for energy rather than increase the supply of energy?

Yeah. It's a complete red herring. Just try to cut your power consumption by one or two toasters. That's the max you could do without starting to feel like you live in a Third World country. And at the other end of that, we're at less than a toaster trying to go up, so we have to provide for these people. And you can't overstate how important it is. What are people fighting over? What are all those wars about? It's access to energy. It's control of those resources, mostly oil. If you could provide a death ray from space full of clean, cheap energy to people, then what are they going to fight about? Dumb shit on Twitter like Americans do.

Can you explain your theory about nuclear energy?

I worked at a lab called Intellectual Ventures years ago, and we invented the most advanced nuclear reactor, called TerraPower. If you see Bill Gates talking about nuclear reactors, that's the one.

In about 2007, and every year since then, we've been unable to get the U.S. government to approve it to build a test core. Nothing wrong with the reactor—it's just the U.S. government has no way of approving any new reactor technology, and the only thing they could possibly approve is old reactor technology. I have a lot of scar tissue from that, because I've been trying to convince people that nuclear reactors are coming for almost 20 years, and I've been wrong all those years.

But this last year, things really radically changed. Last year a huge bipartisan bill was passed called ADVANCE to push for developing nuclear reactors in the U.S.

Last month, Trump signed like four executive orders to push deploying nuclear reactors. And all the hyperscalers woke up and realized, "Oh, to power all these chips we got from Nvidia, we're going to need some nuclear reactors." I think the world has radically shifted in the last year.

In the book, you talk about how nuclear power and nuclear weapons got conflated. Also, if you talk about nuclear power plants, people think of Three Mile Island and the film The China Syndrome. And you throw some shade on the "No Nukes" concerts of the late 1970s. There are no positive examples of nuclear reactors in popular culture.

Actually, 13 percent of our energy comes from nuclear reactors in this country. That's [a] quite positive example. You don't know anybody who lost their life to nuclear reactors. You do know people who lost their life at least early from the pollution from burning coal and gas.

Humans are story-powered creatures—and when we get a story in our head, it controls us. The story we got in our head was we mixed up nuclear reactors and nuclear bombs. We outlawed the wrong one. If we'd done it the other way around, you never would've heard of global warming. That's what's possible, and we're still not being honest about that. It's time to get a new story.

The cool thing is my kid thinks Chernobyl is a TV show for old people. You have to wait out the story for generations sometimes. But we're at that point now. I kind of want to see if I could get Bob Dylan to help us do a "Go Nukes" concert.

"Go Nukes" would be awesome. Do you think there's an argument that even if we had cheap, nonpolluting energy, we just shouldn't consume that much of it because it's wrong to be rapacious or gluttonous in our energy production?

Yeah, but there are other things we do where we're rapacious and gluttonous for generations, and then we learn to get it under control. I think you have to think about a lot of these technology adoption cycles as a life cycle. We're in maybe junior high with social media—still pretty poorly behaved. We're barely in preschool with AI. But for some things, like email, we pretty much got under control, not to mention fire and knives and all the other shit you could really do some damage with.

These things—we're impatient, but you have to learn to get them under control. Everybody overdoes it a little bit with drugs and alcohol, and most people sort of figure out a functional relationship with it. And then there's some collateral damage along the way.

Can you talk a little bit about your vision for AI?

I think of AIs as computational models. Your brain is running those simulations, and now we can make those models in the computer. We can simulate our world better and better and better.

We did this in our lab for epidemiology, and that's the thing that inspires me the most. We started this 15 years ago, advising half of the countries in the world on how to optimally deploy their vaccination resources and epidemiological interventions. In the first Ebola outbreak, 12,000 lives were lost. In the second Ebola outbreak, a few years later, only 12 lives were lost, in part because, using these models, we can get the better answer about how to contain that disease before it spreads. That's what would have been possible with COVID, but we weren't trying.

Why weren't we trying?