What rebuilding AlphaGo teaches us about self-play, RL, and future of LLMs [video] | Dark Hacker News