Hacker News – Telegram
Why traditional reinforcement learning will probably not yield AGI [pdf]
Article, Comments