LLMs work best when the user defines their acceptance criteria first
blog.katanaquant.com
LLMs work best when the user defines their acceptance criteria first
1–10 of 460 posts
Re: LLMs work best when the user defines their acceptance criteria first
#2I tried to make Claude Code, Sonnet 4.6, write a program that draws a fleur-de-lis.
No exaggeration it floundered for an hour before it started to look right.
It's really not good at tasks it has not seen before.
Re: LLMs work best when the user defines their acceptance criteria first
#3Yes plausible text prediction is exactly what it is. However, I wonder if the author included benchmarking in their prompt. It's not exactly fair to keep hidden requirements.
Re: LLMs work best when the user defines their acceptance criteria first
#4I tried to make Claude Code, Sonnet 4.6, write a program that draws a fleur-de-lis. No exaggeration it floundered for an hour before it started to look right. It's really not good at tasks it has not seen before.
Have you tried describing to Claude what it is? The more the detail the better the result. At some point it does become easier to just do it yourself.
Re: LLMs work best when the user defines their acceptance criteria first
#5I tried to make Claude Code, Sonnet 4.6, write a program that draws a fleur-de-lis. No exaggeration it floundered for an hour before it started to look right. It's really not good at tasks it has not seen before.
Have you tried describing to Claude what it is? The more the detail the better the result. At some point it does become easier to just do it yourself.
Most people just forget to tell it "make it quick" and "make no mistake"
Re: LLMs work best when the user defines their acceptance criteria first
#6Most humans also write plausible code.
Re: LLMs work best when the user defines their acceptance criteria first
#7I tried to make Claude Code, Sonnet 4.6, write a program that draws a fleur-de-lis. No exaggeration it floundered for an hour before it started to look right. It's really not good at tasks it has not seen before.
Have you tried describing to Claude what it is? The more the detail the better the result. At some point it does become easier to just do it yourself.
It knows what it is, it's a very well known symbol. But translating that knowledge to code is something else.
Interesting shortcoming, really shows how weak the reasoning is.
Re: LLMs work best when the user defines their acceptance criteria first
#8Earlier quoted context omitted.
Have you tried describing to Claude what it is? The more the detail the better the result. At some point it does become easier to just do it yourself.
Most people just forget to tell it "make it quick" and "make no mistake"
I’m unable to determine if you’re missing /s or not.
Re: LLMs work best when the user defines their acceptance criteria first
#9I tried to make Claude Code, Sonnet 4.6, write a program that draws a fleur-de-lis. No exaggeration it floundered for an hour before it started to look right. It's really not good at tasks it has not seen before.
LLMs are really bad at anything visual, as demonstrated by pelicans riding bicycles, or Claude Plays Pokémon.
Opus would probably do better though.
Re: LLMs work best when the user defines their acceptance criteria first
#10Enterprise customers don't buy correct code, they buy plausible code.