i agree with your general point but i dislike the vapid "count N letters in this phrase" bench that people use to dismiss LLM capabilities

it's like if i was trying to use a screwdriver to hammer in a nail; it's the wrong way to use the tool!

April 10, 2026 - 21:34 UTC

2
0
4
I sort of agree except that the lack of reasoning capability demonstrated is actually relevant to getting correct answers in more realistic cases
2
0
8
What?!
1
0
0
Yeah - I mean it's also just really bad to put the worst version of your product in front of the most eyes.
0
0
1
better models (esp with thinking) can get the correct answer

but the better way is to leverage the LLM to create a deterministic program that does the job
1
0
1
that’s true in a general purpose situation but like, you also don’t want to give that to the AI Overview session, and you would like it to answer this.
0
0
3