Techno-optimism for 2024
What you should be excited about
For those who recently started reading Noahpinion, I created it as an explicitly techno-optimist blog back in late 2020. The success of Covid vaccines and the explosion of renewable energy gave me hope that the conventional wisdom of the 2010s was wrong, and that technology really can solve a lot of our problems after all.
The years since then have only reinforced that feeling. Since I first started to believe in the Roaring 20s, we’ve gotten at least two huge technological surprises: absolutely stunning progress in generative AI, and the advent of the first really effective anti-obesity drugs in history. Those breakthroughs seemed to come out of left field, kind of like mRNA vaccines did, but this is an illusion; in all three cases, years of hard work and incremental progress suddenly reached a point where the technology was ready for prime-time. Which means that there might be a few more such surprises in store for us in the years to come; the decade has six years to go!
Why are all these breakthroughs coming in the same decade? I think some of it has to do with the productivity slowdown of 2005. When the burst of productivity from computerization and the early internet petered out, some private capital and government grants focused on incremental refinements, but some went in search of long-term breakthroughs that would reignite rapid progress. Now I suspect many of those efforts are bearing fruit. I think another factor is that the economy has running hot for most of the last decade, which gives companies both the spare cash and the incentive (high labor costs) to spend on innovation.
In 2021 and 2022 it seemed like productivity was flatlining, which raised a lot of doubts about the idea of a Roaring 20s. But in 2023 there seems to have been a re-acceleration:
Anyway, it’s time for my annual Techno-Optimism post, where I hype up some areas of technology that I’m excited about. This isn’t an exhaustive list, of course, but it hits the main highlights. I put a list of the previous years’ roundups at the bottom, so you can check and see what I mentioned in previous years, and decide whether my excitement was justified.
(Financial disclosure: I have invested in several startups related to the technologies in this post, including a battery-powered appliance company, a company that connects AI companies to cloud providers, an open-source AI startup, and a company that makes robots for drug testing. I would like to invest in a startup making consumer robots, if I find a good one!)
Robot World is coming
A lot of the people who lamented the seeming incrementality of progress in the 2010s would complain that we weren’t getting the future that had been envisioned in the 1960s cartoon The Jetsons. Usually this referred to the fact that we don’t have flying cars. But it’s important to remember that most of the technologies depicted in The Jetsons were actually robots of one sort or another — the most famous being the robot maid that cleaned the family’s house.
This technology was arguably more transformative and utopian than flying cars. Surveys show that people in most countries spend an average of 2 to 3 hours a day on housework and shopping — maybe an eighth of our waking life. And in addition to being unpaid, it’s pretty much our least favorite way to spend our time:
This is why The Jetsons focused on robots doing housework; they knew that this would be one of the technological innovations that would have the most immediate and unambiguously positive impact on Americans’ quality of life — especially women, who at that time were even more disproportionately burdened with housework than they are now. Housework is a job we want the robots to take.
It’s also no coincidence that the first robot that achieved widespread success in the consumer market was the humble Roomba, a disc-shaped automatic vacuum cleaner. But most household chores — washing dishes, doing laundry, and so on — have remained too difficult for robots.
That may now be changing. A Stanford/Google team recently released demo video of its new robot, the Mobile ALOHA, which does household tasks:
Basically, you walk the robot through the tasks about 50 times, and then it can do them itself from then on. There are many other similar multipurpose robot servants in the works. Here’s one being created at NYU, here’s one from Israel, here’s one from Dyson, and here’s yet another from Google. I have no idea who’s going to get there first, but I’m increasingly confident that someone’s going to get there soonish.
I don’t think I’m ready to buy myself a robot servant just yet, but this is far closer to The Jetsons’ Rosie the Robot than anything that’s come before. Robots like this are downstream of two fundamental recent breakthroughs — modern machine learning, and improved batteries. Global venture funding for robotics startups has fallen from its 2021 peak, but those fundamental technological drivers remain, and the potential for a huge consumer market is obvious. And the coming of LLMs may be a game-changer, because it may allow robots to understand voice commands in a very flexible way (a whole bunch of teams are working on this, of course).
Is it too optimistic to think that by the end of this decade, robots will have started to free a nontrivial percent of people in developed countries from the drudgery of housework?
AI art is getting absolutely amazing
Much of the excitement around AI has focused on LLMs, probably because we tend to use conversability as our standard measure of general intelligence. LLMs absolutely exploded in quality in 2022 with the release of ChatGPT, and 2023 saw another substantial improvement in the form of GPT-4. I’m not sure if that pace of innovation will be sustained in 2024 — Dwarkesh Patel has a good post covering the (still unresolved) debate about whether synthetic data will make up for the fact that LLMs have now basically been trained on all the high-quality human writing ever produced. If that doesn’t pan out, we could see a slowdown in the improvement of the fundamental models; at that point, innovative effort would probably shift towards applications like psychotherapy or agents to help you do your office work.
But even more than LLMs, I’m excited by the progress in AI art. At the same time ChatGPT was making waves, AI art applications like Midjourney, Stable Diffusion, and DALL-E were improving by leaps and bounds. In just a year, they’ve gone from making twisted horror-doodles with too many fingers to producing startlingly photorealistic images of things that no human has ever drawn or perhaps even imagined before:
And for fantasy illustration, they’re as good as almost any human artist alive.
The reason I’m even more optimistic about AI art than about LLMs is the availability of training data. Villalobos et al. (2022) show that high-quality language data will be exhausted in the early to mid-2020s (assuming no synthetic data), but that image data won’t be exhausted until the 2040s:
And new image data, at least of some kinds, is easier to generate than high-quality human-generated language data — just send a bunch of drones out to go take trillions of photos.
So as good as AI art is getting, I think it has the potential to keep getting better at an astounding rate. Already, focus is shifting from still images to video generation. Obviously this is going to create some problems for our information ecosystem, as deepfake videos proliferate — some people think deepfakes may already have swung an election in Slovakia. But putting realistic video generation at the touch of everyone’s fingertips will be absolutely life-changing. (I’ll probably use it to create movies of giant rabbit-dragons.)
The Decade of the Battery powers onward
One technology that I’m consistently hyper-optimistic about is batteries. This is because batteries represent something humanity has never really had before — a way to store a large amount of energy and carry it around.
The storage aspect of this, just on its own, is incredibly important. Increasingly, people are using batteries as the solution to smooth out the intermittency of solar and wind power, providing stable electricity supply through night, rain, and even winter. Around the world, battery storage for grid power is growing exponentially. The EIA expects it to nearly double in the U.S.:
Texas, the U.S. state that builds the most renewables despite its conservative rhetoric, is adding massive amounts of batteries.
Even before we consider portability, the ability to store energy has tons of applications other than smoothing out renewable electricity. For example, Impulse Labs, the battery-powered appliance company that I invested in, has just released its first product — a super-powered stove — and is now taking preorders.
(They’re also hiring.)
But the importance of batteries goes beyond storage; batteries also make energy portable, meaning they compete with liquid fuels like diesel and gasoline. Bloomberg New Energy Finance now predicts that by the end of the 2020s, electric vehicles will represent over half of new U.S. passenger car sales. Globally the transition will also be rapid, thanks in large part to China, which has ramped up its EV production enormously and become the world’s largest car exporter. That’s causing trade friction, but ultimately it’ll be good for both the environment and for consumer satisfaction.
Batteries are also behind the aforementioned incipient robot revolution, of course. And they’re also transforming the skies. Drones are proving themselves to be the essential weapon of modern warfare (as I predicted long ago), and small battery-powered quadcopter drones are a key part of that. Of course, drones will also be useful in various peaceful applications, but the revolution in military affairs seems like it’s the most important at this point.
And the most amazing thing about batteries is that the fundamental underlying technology is far from mature. Two huge innovations are just now making it to the commercialization stage: sodium-ion batteries and solid-state batteries.
Solid-state batteries are just better in every way than the batteries we use now — they’re lighter, smaller, stronger, and safer, they have greater range, they recharge more quickly, and they’re better for the environment. Toyota is going to begin selling cars with solid-state batteries soon; we’ll see how that goes, but my guess is that someone will commercialize solid-state batteries successfully within the decade.
Sodium-ion batteries, meanwhile, have a lot of performance advantages and use a lot fewer scarce metals than lithium-ion batteries. Chinese companies are now rolling them out for grid storage and possibly for EVs.
So as amazing as the battery revolution is right now, it’s just getting started.
Biotech is still booming
We usually think of biotech as distinct from innovation in electrical and mechanical applications, but the two are beginning to merge. Computation is essential to modern drug discovery and chemistry in general, and AI looks like it’ll be another important tool. Meanwhile, biological experimentation is increasingly making use of robots in the lab.
But the biotech revolution that’s unfolding right now isn’t just downstream of those other technologies — it’s the result of decades of massive work and deep-pocketed investment that’s now coming to fruition.
MRNA vaccines, the heroes of the Covid pandemic, are now being applied to a ton of other diseases. A new mRNA vaccine for malaria promises to help conquer humanity’s most deadly chronic illness. And a new mRNA vaccine for pancreatic cancer — which you actually take as a therapy after you’re diagnosed, not before — promises to save a lot of lives from one of the deadliest cancers. Other cancers are probably also treatable this way.
Another technology that’s just now coming to fruition is genetic engineering, especially via CRISPR. The FDA just approved a drug that actually goes in and reengineers the cells in a living human body to treat sickle cell anemia. Other such drugs are in the pipeline. Genetic engineering is also proving useful for modifying bacteria and other organisms for useful purposes, such as producing fuel or fighting cancer. Of course all of this is in addition to the potential for reengineering the human species itself, but that’s still on the horizon.
A third major area of technology whose potential is still on the horizon, but could be realized in the near future, is synthetic biology. This year, scientists synthesized the DNA to make an artificial yeast cell. Synthetic biology could eventually be used to create all kinds of custom-built organisms to do various jobs.
I should also give a shout-out to my friends at the company Loyal, who are close to getting FDA approval for the first commercial longevity drug for dogs. I hope the drugs are useful for pet rabbits, or that they make a rabbit version.
But the most important biotech discovery of the year sort of came out of left field. A class of drugs known as GLP-1 agonists, including Ozempic, Wegovy, and Mounjaro, long used to treat diabetes, has proven to be incredibly effective at helping people lose weight. The obesity epidemic is America’s #1 health problem, and is spreading to the rest of the world. These drugs are the first really effective pharmaceutical approach to combatting that epidemic.
And on top of that, one of the ways that these drugs treat obesity — basically by reducing food cravings — also makes them useful against a variety of other addictive behavior, like alcoholism and smoking. This makes them potential game-changers for many of the ills that afflict advanced societies. Already, demand for the drugs has supercharged the economy of Denmark, home of the company that makes Ozempic and Wegovy, due to massive demand.
So anyway, the promise of biotech really seems to be coming to fruition this decade, in a variety of very awesome ways.
The reconquest of space
One of the most important inventions of the 2020s that hasn’t worked yet is SpaceX’s Starship, a gigantic reusable rocket bigger than the Saturn V, which looks like a 1930s comic book rocket-ship, and can be made cheaply and in large numbers. Starship hasn’t successfully flown yet, but each test is getting closer. It’s pretty majestic to watch:
Starship seems extremely likely to become usable within the decade, after which a whole lot of space stuff will become viable that’s currently impossible — asteroid mining, space manufacturing, Mars bases, and so on. The fact that the rocket can also hop up and land elsewhere on Earth very quickly makes it potentially extremely useful as a rapid mode of transportation. More than any other technology, Starship is promising to take humans back to space, after our long post-Apollo retreat. This time, we’ll be there to stay.
Previous annual techno-optimism posts
Here are the versions of this post for past years, so you can see whether my optimism panned out. So far I think very few of my hopes have been dashed.
Big fan of your work Noah. Your comments on AI art made me think about some of the angst that was floating around the webs last year - the meme of robots becoming the poets, authors, and artists, while humans were relegated back to working in the fields - I wonder if art as a business will evolve in the same way that the music industry did, i.e. with less and less revenue over time from media sales and digital streaming, growth in importance of live touring, and the emergent dominance of mega-artists and mega-tours. You could see some parallels - if digital art was largely commoditized with close to zero residual commercial value, perhaps we'll see more growth of galleries and exhibition spaces for physical and performance artworks over time?
One other gamechanger that could come out of robotics: robot nannies and childcare workers. Given the cost of childcare, I'm genuinely convinced that their development would solve our birth rate crisis within a few years.
Also, thank you for pointing out the synergy between advancements in biology and in computing. CRIPSR gets the press (reasonably so) but it only solved the problem of how to make cheap and rapid targeted gene edits. The true revolution in biology will come about due to the advancements in -omics and bioinformatics we've been seeing the past decade, allowing us to accurately measure and model cell-wide changes in phenotype at the molecular level.