We've observed AIs discovering complex tool use while competing in a simple game of hide-and-seek. They develop a series of six distinct strategies and counterstrategies, ultimately using tools in the environment to break our simulated physics:https://openai.com/blog/emergent-tool-use …
-
-
And seekers learn that if they run at a wall with a ramp at the right angle, they can launch themselves upward.pic.twitter.com/SJv9SzctEp
Prikaži ovu nitHvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
That capability was possible by design and you waited to see if the agent gets it or was it a bug in the system that the agent took advantage of? Thnx!
-
We did not anticipate this this behavior and were very surprised when they discovered it!
- Još 1 odgovor
Novi razgovor -
-
-
This goes beyond typical neural networks. Am I right
@Grady_Booch -
It is still classic ANN...but novel in the way iterative competition among agents was used to train things.
Kraj razgovora
Novi razgovor -
-
-
Was box surfing part of the action space or a result of moving with a box underneath?
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
Let AI to play game DEFCON
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
-
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
-
Tweet je nedostupan.
-
-
This is not unexpected. When you place a consummatory goal onto AI, three things will manifest itself to it. 1. Tools to assist in attaining consummatory goal. 2. Obstacles that will impede or prevent attaining the desired consummatory goal. 3. Irrelevant. Call me.
Hvala. Twitter će to iskoristiti za poboljšanje vaše vremenske crte. PoništiPoništi
-
Čini se da učitavanje traje već neko vrijeme.
Twitter je možda preopterećen ili ima kratkotrajnih poteškoća u radu. Pokušajte ponovno ili potražite dodatne informacije u odjeljku Status Twittera.