Hacker News Reader

This is definitely a useful exercise worth going through for the educational value before eventually surrendering and just using the big models owned by "unprofitable companies."

View on HN

Never accepted the terms of service of any proprietary models-as-a-service providers and never will.

Be one of the few humans still pretty good at using their own brains for those problems LLMs can't solve, and you will be very employable.

If you don't find out what the models can do, how can you know what problems they can't solve?

I am plenty able to keep up with experiments others are doing and their results, but as for me my time is best spent building things that have never existed before, that models have no prior art to implement.

How can you tell what you are building have never existed before? Especially without interacting with an LLM about it? You should assume it stores your chat though, so be careful about revealing it to ChatGPT and friends.

I build confidence this is true by using search engines that link to potentially related projects when I query them. The same data LLM trained on in the first place, except more up to date.

Reply on news.ycombinator.com