[anonylizard]

Deepmind is trying to combine "Alphazero" style algorithms with LLMs to build Gemini.

[worrycue]

We will see if they succeed and create a better AI that doesn’t relies solely on memorising answers.