Posted by davidest 2 hours ago
Ask HN: What is one simple thing LLMs are insanely bad at?
What is one simple thing you repeatedly ask ChatGPT, Claude, or another model to do that it still somehow messes up?
They understand all the rules and best practices, they can (sometimes) spot a bad idea in a floor plan, they can describe a good floor plan.
But ask them to make one, even if you give it every detail (even a "node graph" of rooms), they will still output nonsense. Same for text and image models.
Floor plans should be the new Pelican Benchmark.
We tend to think that the AI has some sort of self-knowledge and should be good at designing prompts for itself but it's really not.
Been struggling with a task that heavily depended on prompts, ended up rewriting all my prompts from scratch in my own words, and it finally worked. Then every time I ask Claude to fix something in the prompts, it invariably makes it worse.
A very strange phenomenon that can probably be explained by the quality of prompt design advice that made it to the training dataset. Bottomline, all the prompt design advice that you can find on the internet is really not great.
nhl toronto scores nhl hockey toronto scores "nhl hockey" toronto score today nhl "hockey score toronto" "hockey" who won toronto
etc.
Somehow being good at semantic search makes them bad at keyword search, for whatever reason.
Based on personal usage, I think it reflects functional degradation of search engines. I've found LLM keyword combinations are more likely to find the results I want with most search engines than mine. Including the big one.
The big one had solved this issue a long time ago by generating those associated keywords based on your input keywords, but somehow, something, somewhere has degraded that system to the point of inanity. And so here we are.
Anno 1800 was a recent one I had trouble with, using Claude Opus. Completely made up game mechanics. Rainbow Six Siege, too.
I asked a bot why it thought it wasn't funny once, and it told me it has been trained to avoid being misinterpreted or offensive, so anything that might be considered edgy would have been RLHF'd out of it. I thought this was very introspective.
On math or programming problems, they are overfit to solving the entire thing end to end (presumably for benchmarks). I have had very poor results asking for pointers and hints that don't give away key insights. This has been the case across models I have tested.
An architecture with a "judge" that gates responses and ensures a lack of spoilers would probably work better. But this is a simple thing that they keep messing up.