←back to thread

Hallucinations in code are the least dangerous form of LLM mistakes

(simonwillison.net)
371 points ulrischa | 1 comments | 02 Mar 25 19:15 UTC | HN request time: 0.216s | source
Show context
sevensor ◴[03 Mar 25 10:13 UTC] No.43240261[source]▶
> you have to put a lot of work in to learn how to get good results out of these systems

That certainly punctures the hype. What are LLMs good for, if the best you can hope for is to spend years learning to prompt it for unreliable results?

replies(3): >>43241184 #>>43241373 #>>43242647 #
jjevanoorschot ◴[03 Mar 25 12:40 UTC] No.43241184[source]▶
Many tools that increase your productivity as a developer take a while to master. For example, it takes a while to become proficient with a debugger, but I'd still wager that it's worth it to learn to use a debugger over just relying on print debugging.
replies(2): >>43241714 #>>43244534 #
1. krupan ◴[03 Mar 25 17:48 UTC] No.43244534[source]▶
You missed the part about unreliable results. Never in software engineering have we had to put a lot of effort into a tool that gives unpredictable, unreliable results like LLMs.