The AI Hype Cycle: Are We Putting the Cart Before the Horse?

I’m getting old. Most people hate getting old, I sort of love it. Yeah, I can’t drink like a horse any more, but I have so much more history and understanding of the world.

Part of the understanding is finding the hype cycle. Roughly put, the hype cycle has 5 stages: Innovation Trigger, Peak of Inflated Expectations, Trough of Disillusionment, Slope of Enlightenment, and Plateau of Productivity. The hype cycle is not a guarantee of success or failure. Sometimes the product just dies in the Trough of Disillusionment.

A few examples of hype cycles of the past include cryptocurrency, NFTs, virtual reality, 3D Printing, internet of things, web3, the metaverse, etc… When you step back and look at these, the reality is that none of these had wide spread use cases, but instead had massive hype behind them.

The AI Hype Cycle

I don’t know quite where we are in the hype cycle, but it’s likely somewhere in the first two stages. Google recently just moved it’s stock price $150B just by saying AI a lot of times on stage.

I’m not saying that AI won’t be a huge development in the future, but the reality is these models are far closer to the cool than useful end of the spectrum. 10 years from now will likely be a very different deal, but when you follow hype cycles it is more about what happens now.

AI Strikes Back

While I do love getting old, I’m not some boomer who poopoos any new technology. Much of the argument that I’m making against AI could apply to the internet in 1998. 25 years from now, comparing AI to a fax machine would be foolish.

My point is more that are we all sure that two years from now we won’t be thinking that we sort of put the cart before the horse? Generative AI is the beginning of something, but I think people are acting like it’s a finished product when it’s clearly not.My point is more that are we all sure that two years from now we won’t be thinking that we sort of put the cart before the horse? Generative AI is the beginning of something, but I think people are acting like it’s a finished product when it’s clearly not.

I’ve heard a lot of “we haven’t seen this pace of innovation in a long time”. But to be honest – have we really seen that much innovation or just that much replication? A lot of companies are doing the exact same things. Yeah, chatGPT4 is better, but is a new model every 3 years really a crazy pace of innovation? It is great to be open source competing with the tech behemoths. It is fun to play with image generation. Large language models are better than search on 10% of questions. But we are 5 years out (at least) from mass impact of AI. The reality is that the market doesn’t have that kind of patience though, so a correction will come on the AI hype. That’s how the hype cycle works.

P.S the beauty of a blog is I have record of myself being an idiot when 1 year from now our AI overlords are ruling, I can at least learn from my mistakes!