Yes, RL is mostly about learning from rewards (Reward is Enough - https://www.deepmind.com/publications/reward-is-enough https://www.deepmind.com/publications/

Yes, RL is mostly about learning from rewards (Reward is Enough - https://www.deepmind.com/publications/reward-is-enough https://www.deepmind.com/publications/reward-is-enough) and GPT-3 has been used so far only to augment agents, not to do reward based learning. The current obstacle is speed - it needs to run in real time. Maybe we need more efficient algoritms, maybe we'll have better hardware. It will come soon enough.