DeepMind invested heavily in pursuing RL as the holy grail for AGI. Which still seems like a plausible approach to me, we're not there yet! But despite inventin

DeepMind invested heavily in pursuing RL as the holy grail for AGI. Which still seems like a plausible approach to me, we're not there yet! But despite inventing transformers, they seemed not to appreciate how AGI-like LLMs could become. And now OpenAI has beaten them to the punch, by actually releasing a product, which is not Google's forté.