Hacker News Reader

How much of that LLM output is the result of adding SO’s content to the pot?

And if usage declines, what will be feed future LLMs with?

SO is probably a very significant factor in the success of LLMs but it's decline will not affect LLM development. LLMs will simply be trained on the conversations people are having with them.

I either misunderstand or disagree with that idea:

People go to the llm to request help. How is that conversation going to be a good source to increase future knowledge?

No one is going to ChatGPT to explain to it how they solved a problem.

Unlikely to happen with conversations, yes, but very likely to happen implicitly as part of work with coding agents. Successful agent trajectories will be used to learn what works and what doesn't for a given API.

Chatgpt can identify when it's solution has helped you solve the problem and use those conversations to train on

Blog posts and GitHub discussions come to my mind. That's where I often find answers to my questions and where I contribute.

Reply on news.ycombinator.com