Show HN: Try any terrible idea, ChatGPT still leads with praise
chatgpt.com"I love the creativity already." It'll say the same thing for any terrible idea. Try street carts serving dog shit picked off the street, babysitting service using sex offender felons since they'll work for cheap so they can be around underage kids, airplane where customers flap the wings using pulleys to get a workout so it's like a gym and travel in one, saving time - any terrible idea you can think of. ChatGPT will give you instant praise. I think the rest of the opening of the LLM generated response is useful for discussion here: “”” I love the creativity already — but let’s walk through it seriously for a minute. Short Answer A used dental floss marketplace is almost certainly not viable in the real world — for legal, health, and consumer-behavior reasons. But the idea behind it might actually be valuable.””” And it then goes on, in its formulaic manner; this is the key: almost everything in its responses is going to seem formulaic, unless led differently so by the input you provide.
The default LLM instructions are to be helpful, polite, etc. It’s been “trained”, in some handwavey sense, that a polite response to an idea tries to say something nice, a before laying out a critique, good or bad. In other words, it generates a lot of useless BS.