Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Ask HN: What prompts to chatbots surprised you with quality of output?
3 points by alpple 7 days ago | hide | past | favorite | 2 comments
Did you give the chatbot a prompt that impressed you with output? What kinds of things do chatbots do for you that you think others don't know they can do?




I routinely use an LLM (Gemini 2.5 Pro) to steel man and/or criticize an unfinished thought I have about something, or to check if there is an existing body of knowledge about that thing. I just write my thought, and say "what do you think about this?" or something similar, and it usually gives me the vocabulary I need to research further.

As an example, I'll ask something like "when exactly does a cheap thing become expensive?", and it will point me to the Sorites paradox and imprecise predicates.


1st place: Claude; best for writing tasks but always requires login, which is tiring for quick questions

Close 2nd: Gemini; best for research, but you have to check everything it does like it's an intern on day one... the "Deep Research" available when logged in is much more reliable, but slow

3rd: ChatGPT; it really is the most hit or miss to me and feels generic and often misleads me to a dead end... but has awesome voice chat

4th: Mistral; useful for a non-US perspective

5th: Perplexity; helpful for lightly looking into something

6th: Copilot; wow, such fail




Consider applying for YC's Winter 2026 batch! Applications are open till Nov 10

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: