One thing I've learned is, there's a ton of interesting stuff in the world. Super talented people. But, what surface and gets widespread recognition and adoption is often very much determined by luck, and these days by algorithms. It's very possible there's some really cool AI games out there already, and they just haven't been discovered.

The other aspect is that a lot of people in the AI community are completely hooked on the reward-seeking of 'new thing'. Every hour there's some new tool to try out that 'changes everything'. But, without grinding it out you might never create something interesting by constantly jumping to the next new thing. Because, technological capabilities are just one part of the story.

Third, what makes something 'fun' often requires tons of iteration and a lot of LLM-type tools still rely on snapshots of applications. They're not fully there yet in terms of what happens in between moments. There are lots of principles they can draw on from and get a good prototype quickly, but it again comes back to really just slogging through lots of iterations until you find something fun.

Finally, I think with every new medium you get a first wave that really just copies the medium that came before it. Early radio was a lot like theater. Early TV was a lot like radio. I think eventually we'll see people – probably younger people – who are fully into AI and will find interesting things about that medium to express in ideas, and we'll get some really new and interesting stuff as a result.