4 Comments
User's avatar
Antoine Dusséaux's avatar

I agree with the overall thesis except for "The Death of User-Generated Content". I also ditched Google for ChatGPT and... Reddit (and some other high-quality public forums). LLMs are great for general current knowledge, but new theories, assumptions or ideas thrive on social media: that's where innovation is. For now, LLMs can't innovate much.

Nicolas Bustamante's avatar

My bet is that *this year* LLM will innovate. E.g solving complex and novel mathematical problems humans can't.

Antoine Dusséaux's avatar

Yes in maths but less so in real world "innovation" such as "there's this really good hidden café in that street of Baku that just opened" or "More and more people are switching to this ideology in the countryside of Kazakhstan". There are things that I discovered on Reddit that I couldn't find anywhere else and that LLMs couldn't find either (unless by scraping Reddit). For sure those things aren't super valuable (compared to a math theorem) but that's the small frontier where humans still have some "expertise".

James's avatar

I was thinking about this too Nick. It’s easy to get spooked about. I want say I’m not!