The Things That AI Won't Change

Where to place bets on stability

Happy Sunday and welcome to Investing in AI. I’m Rob May and I run the AI Innovator’s Community in NYC and Boston.

For my topic this week, I’ve been inspired by Morgan Housel’s book Same As Ever, which looks at the things that don’t change over time even as technology advances. It inspired me to think about AI through that lens. So much is going to change as AI rolls out across industries. But, what won’t change? What things are either outside the impact of AI or, even if AI can impact them, get rejected in favor of non-AI approaches?

One area I see is the idea of human relationships. Summarization is a constant theme of AI and that includes summarization of human relationships. There is an emerging trend of bots that respond to things for you, including other people. But if your bot talks to me to keep me updated on your life, but then I deploy my bot to chat with you and your bot, and our bots talk to each other, are we really building a relationship? Are summarizations of key life milestones without the actual interaction the same?

To take an extreme example - a summary of a sexual experience isn’t the same as the sexual experience itself, and in many cases of human relationships, the situation means a lot more than just the basic transaction of information. I generally don’t invest in AI that tries to summarize or replicate human experiences.

A second area where AI will have little impact is areas where there is no standard amongst humans. For example, I see a lot of startups that want to use AI to validate truth online. But if humans can’t do it, how can AI? A significant part of the population still believes there was fraud in the 2020 U.S. Presidential election. Having an AI tell them there wasn’t isn’t going to change their point of view. While there are things AI can do to help in this space, like finding and tagging AI-generated content, or monitoring and aggregating information about what’s going on in the news, it boils down to this - when humans can’t agree on truth, AI can’t help.

The third and final area I’ll suggest as an area AI won’t change things is areas that are regarded as status symbols. I wrote about this in 2019 in What Horses, Watches, And Bookstores Can Teach Us About Why Automation Won’t Kill Jobs. Here is an example from that post:

Two hundred fifty years ago, everyone owned a horse. That’s how you got somewhere. When cars came around, it didn’t kill off the horse industry altogether. Instead, horses became expensive status symbols. To own a racehorse, or to participate in equestrian events, it it’s own culture and community filled with mostly wealthy people.

In situations where something is, or may become, a status symbol, there may be limited chances to apply AI. AI works best when the goal is automation, or lower cost predictions, or something like that. Status symbols don’t usually benefit from either of those use cases.

It’s common to extrapolate new technology to everything. But it rarely works out that a new tech upends every area of the economy. Charlie Munger says then when analyzing something it’s important to “invert, always invert.” One way to think about where AI goes and the impact it has is to think about the areas that might stay the same even in the face of AI advancements. If you have areas you’ve thought about that AI can’t touch, I’d love to hear them.

Thanks for reading.