Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
Invictus0
40 days ago
|
parent
|
context
|
favorite
| on:
OpenAI claims gold-medal performance at IMO 2025
The "AI" that won Go was Monte Carlo tree search on a neural net "memory" of the outcome of millions of previous games; this is a LLM solving open ended problems. The tasks are hardly even comparable.
yobbo
40 days ago
|
next
[–]
A "reasoning LLM" might not be conceptually far from MCTS.
SonOfLilit
40 days ago
|
prev
|
next
[–]
I really don't like the use of the word memory here, even in quotes. AlphaGo has a much better "understanding" of Go positions than mine (7k).
gafferongames
40 days ago
|
prev
|
next
[–]
And then they created AlphaGo Zero, which is not trained on any previous games, and it was even stronger!
https://deepmind.google/discover/blog/alphago-zero-starting-...
yorwba
40 days ago
|
parent
|
next
[–]
AlphaGo Zero was also trained on millions of games, they just weren't games played by human players.
Workaccount2
40 days ago
|
prev
[–]
Nothing that uses a mathematical model for solving a problem will ever reason because reasoning can only be done by things we don't understand...
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search: