The Internet Boom of the late 1990s laid the groundwork for subsequent booms in social media and cloud computing. The thing I've been thinking about as an investor is... in 2030, how will I finish the sentence "the AI boom of 2015-2019 laid the groundwork for..."??
Bringing "intelligence" to software and hardware is going to change many things. But every new idea is just as dependent on the ecosystem, both social and technical, around it, for success. Many ideas are too early for their time. I once worked as a sales engineer for a company working on a new high speed wireless protocol in the mid 2000s. When we started, wifi was slow as hell. By the time the hardware companies we were working with got the new protocol to work, wifi had gotten surprisingly fast. No longer were we 20x faster. We were just 2x faster, and way more expensive. The ecosystem we started in wanted our product, but the ecosystem we finished in did not.
How do you anticipate what these subsequent booms caused by AI may be? A good tech startup usually has a few parts. First, it makes use of an emerging asset - an asset that not a lot of people understand how to use or how to fully value. For AI, this is more likely to be either labeled data sets, AI hardware, or AI value chains - specifically the interaction between workflows and technology and how they shape each other to create value. I think new algorithms and neural network topologies specifically are probably overplayed, and everyone is competing there, and the marginal value of innovation is lower, economically.
Secondly, you need to be in a tech ecosystem that can support the new idea. For example, having a superior insight about two datasets that could be combined to train a novel model won't do you any good if the data types are hard to work with together for some reason, and the place you work is barely thinking about AI in the first place. One of the really hard things is aligning the incentives in a new ecosystem so that all parts support the new idea. Often different pieces of a tech ecosystem support different forms of whatever is new.
And third, you have to follow the money. Who benefits, and why? Customers either need an economic reason, or a proxy for an economic reason (like social status) to invest in the new idea. Where these three things overlap, you have a good chance of having something successful.
The hardest piece is often understanding the new assets. What happens is, for example, when a new type of chip, targeted towards an AI use case, becomes common, people start to think of other uses for it - things it wasn't designed for. In the case of Web 1.0, once a web browser was standard, people starting thinking differently about interfaces and programming possibilities.
I don't know yet what will drive the second (and third?) booms in AI, but I expect they are coming soon. Intelligence as a concept is going to work it's way through everything eventually, but in what order? And where does it have the biggest impact? If you have ideas I'd love to hear them.
Thanks for reading.