News for hackers. Stuff that matters.

A comment on The 70% problem: Hard truths about AI-assisted coding. Parent comment

Re:The 70% problem: Hard truths about AI-assisted coding by pizza on Friday December 06, 2024 @06:46PM (#42345855)
But actually whenever this happens you get rich signal about the hidden didactic assumptions of your method of prompting it about things you yourself are unsure you can verify for yourself, and also of how you thought the tool worked. This is a good meta skill to hone.
[ Parent | on HN ]
Re:The 70% problem: Hard truths about AI-assisted coding by dartos on Sunday December 08, 2024 @08:45AM (#42357101)
In the same way going to a fortune teller can give me rich insight about my love life and work life balance, sure.
[ Parent | on HN ]
Re:The 70% problem: Hard truths about AI-assisted coding by pizza on Monday December 09, 2024 @12:54AM (#42363382)
A visitor to physicist Niels Bohr's country cottage, noticing a horseshoe hanging on the wall, teased the eminent scientist about this ancient superstition. "Can it be true that you, of all people, believe it will bring you luck?"

"Of course not," replied Bohr, "but I understand it brings you luck whether you believe it or not."

s/luck/lower negative log likelihood. Besides, a participant can still think about and glean truths from their reflections about a conversation they had which contained only false statements.

[ Parent | on HN ]
Re:The 70% problem: Hard truths about AI-assisted coding by dartos on Monday December 09, 2024 @09:13AM (#42366413)
What point are you trying to make?

That it’s better to be superstitious than not, because maybe it matters?

[ Parent | on HN ]
Re:The 70% problem: Hard truths about AI-assisted coding by pizza on Tuesday December 10, 2024 @02:56PM (#42380729)
The internal mechanisms by which a model achieves a low negative log likelihood become irrelevant as it approaches perfect simulation of the true data distribution.
[ Parent | on HN ]
Re:The 70% problem: Hard truths about AI-assisted coding by memhole on Monday December 09, 2024 @09:50AM (#42366695)
This is how I think about it as well. It’s like the equivalent of trying to build an ML model with only positive examples otherwise
[ Parent | on HN ]