Why AI is not the investing opportunity of the decade
Peter Thiel has described the cultural narrative surrounding AI as being broadly dystopian, consisting almost always of negative themes such as unemployment and mortal danger. He’s right — the latest story on the topic was about a guy who created a deep fake of Tom Cruise, then warned how dangerous this technology was¹. This outlook is now pervasive, Janan Ganash put his finger on it recently² in discussing Kazuo Ishiguro’s Klara and the Sun — if such a subtle thinker could be so heavily imbued with techno-pessimism it really says something about the general outlook of society towards AI¹.
But this narrative is limited to AI specifically isn’t it, not technology in general. Right now we are riding a market bubble of historic scale driven by people’s excitement about what is coming over the next decade — electric vehicles, Fintech, crypto. Logging into Clubhouse last night I found myself in a room discussing Elon Musk’s plan to release a techno track as an NFT³. The contrast between the general narrative around AI, and the energy of this room was the difference between night and day.
This room has literally thousands of people in it, and it was buzzing. MC Hammer was there, Disclosure , Pussy Riot spoke for a while. The room was over-flowing with positive energy, optimism and generosity; diving deep into topics not normally part of tech chats — politics and art. It was clubhouse conversation at its best: authentic, respectful, passionate. And it went on for hours. This is a feature I’ve noticed of Crypto chats on Clubhouse they go on for hours and hours, marathon discussions, Cafe Bitcoin is another that often seems like its just ‘always on’. This feature is a powerful tell, I think, for a space which is so overflowing with energy that is doesn’t need timetabling like the more ‘managed’ chats put on by people trying to build an audience on Clubhouse. It is pure, spontaneous organic energy.
If the AI narrative is dystopian, this must come as close as we have anywhere to embodying utopianism. And yes you can point to there being lots of speculation and hunger for price rises, but underneath it all is a very powerful techno-utopian streak that goes far, far beyond simple greed. Is it allowable as evidence that I haven’t stumbled into any Clubhouse rooms on AI yet? Browsing through the clubs they have the same tone as many AI newsletters, which is overwhelmingly dry, functional and bland. If this was freshers week at uni and you were walking around the hall checking out all the clubs and societies, the AI lot would be something like the chess club — very clever yes, technically impressive, but kind of a backwater in social terms. By contrast you would find it impossible to ignore the crypto stalls, brimming with energy, music, like a magnet for social energy.
In some sense this is strange because AI is truly one of the most remarkable technologies we have ever seen on this earth, on several occasions over the past decade I’ve been literally floored by an AI demo showing a new technology for the first time. GPT-3 is the most recent example — you see it in action for the first time and you’re entire perspective about what is possible collapses and is reconfigured in an instant, its like the boundary of human capability just got an upgrade. This technology has exceeded the expectations of many practitioners and continues to do so. If GPT-3 doesn’t do it, try Alpha Go, the Deep Mind programme that beat world champion Lee Sedol hands down at a game many thought computers couldn’t play.
AI is both perhaps the most important technology of our time, and yet from a popular narrative perspective verges between deep pessimism and simply…disinterest. For a technology that is being deployed in every corner of the earth as we speak, the conversation is surprisingly abstract and remote, tending towards themes of geopolitics or general dystopia. There is surprisingly little the other way in fact, talking about how it is working today in a way ordinary people can relate to. With Crypto you have price movements to anchor the popular narrative in something tangible — anyone can buy it; with electric cars you have Tesla, a product everyone can aspire to own. But how does the typical person relate to AI? What is there to relate to?
This I believe is a non-trivial factor in the way the narrative has emerged. The popular impressionistic image of AI is of a cold, aloof, vaguely threatening force — and the word force is significant because it implies something abstract and beyond human control. In many ways these are just by-products of the nature of the technology. As Benedict Evans has said, the best analogy for AI is the database — a general purpose technology — and the signature feature of this is that users don’t know they’re using it, and couldn’t care less how its done. People happily start using AI-powered type-ahead suggestions in their gmail without giving a second thought to how it works; they certainly don’t relate it, I expect, to the swirling, grand-historic, remote narratives of AI that dominate the media.
Because AI doesn’t really have a face does it? It is without personality or brand. There is no symbol, no company, no individual that represents it. But then again, how many people can name a database company? Its the nature of low-level, general-purpose technologies to manifest like this — indirectly. AI is not going to be ‘a company’, it is going to be used (via API probably) by millions of companies the world over to make their products, services or processes 10%, 20%, 500% better. But all of this is unseen, it’s a back-office transformation — a revolution that will not be televised.
Even within the tech industry, where your typical engineer can tell immediately that some product they’re using has incorporated AI,
relatively remote and inaccessible, because you need a lot of data to get started doing anything meaningful, and where does some indie-hacker, some would-be entrepreneur, magic this data from? AI is a force-multiplier, but you still need something to multiply from, not just data but all the other things that make up a company — brand, users, capital, product. AI isn’t a primitive you can work with to create something completely new in its own right, its a tool to amplify something you already have, and this works contrary to entrepreneurialism. This is compounded by the fact that the technology is almost immediately commodity — if you have access to the GPT-3 API, great, but so do 10,000s of other people, and good luck trying to build a better text prediction model that Open AI.
This leads me to think that while AI feels likely to be a once-in-a-lifetime technological revolution, it might also be an oddly silent one; unlike major technological upheavals of the past this isn’t visible — no trains barrelling through the countryside, no memory of the first time you flew in a plane, no app on your home screen. Perhaps this will change somewhat with self-driving cars and delivery robots, but much of it will remain back-office, B2B.
And this is why I don’t like AI very much from an investment perspective, despite it being impossibly compelling from a technological perspective. It doesn’t fundamentally change the value-creation dynamics for many entrepreneurs (there will be many notable exceptions), and it is surrounded by this bad energy and techno-pessimism. As Peter Thiel says “At the end of the day, technology is about people, its not about inanimate forces… the stress is on individual, small teams that start companies, that do new things”⁴. And what is interesting to invest in is where the most ambitious, imaginative, bold people are spending their time, love, energy and reputation; and broadly speaking that just doesn’t really feel like its AI right now.
¹ https://www.thetimes.co.uk/article/deepfake-videos-of-tom-cruise-watched-by-millions-tr8lkmfdk
² https://www.ft.com/content/f6772f08-450e-4490-9e3c-8ed455f19c8d
³ https://www.forbes.com/sites/rachelsandler/2021/03/15/elon-musk-is-selling-a-song-as-nft
