Originally posted on my LinkedIn on 11/25/24 @ 3:18 pm.

As an AI skeptic—which is to say, I’m extremely optimistic about the things AI has proven it does very well, and the things it’s given strong indications it’s moving towards really doing well, but deeply impatient with all the hype about speculative functionality outside of those—this really put a fine point to me on part of the problem. Not with AI, but with what people expect of it, what they spend money expecting it to do.

AI DOESN’T INNOVATE. AI is specifically engineered NOT to innovate. It specifically attempts to create something that matches what it previously ingested. If it deviates from that, innovates, we call that “hallucination”, and it makes AI less useful, not more. This is because AI is a tool.

For a designer to think AI innovations is going to change your designs is comparable to an architect thinking a new drafting technique, or an easier way of creating blueprints, is going to innovate your architectural designs. It’s not.

Current generative AI empowers YOU to create with much less work. If you create innovative things now, you may create them faster. If you do not, you may create uninnovative things faster. It does not create anything. It’s a mechanical pattern-replicating process. (Which, by the way, is why AI output should be copywriteable, but that’s a different conversation.)

That word ‘replicating’ is key.

That’s it. It’s a tool. It helps you do what you already want to do.

The day may come where AI has a deep enough symbolic representation of the development of new ideas that it can extrapolate and replicate something like that process, and hit us with something that seems truly new and useful (as opposed to new and uselessly random, which it currently does absolutely fine.) We’re not there yet. And nothing about the current technology, other than fanciful science fiction thinking inspired by it, shows definitively that that’s coming.