AGI Challenges

Keith discusses the challenge of achieving AGI by programming human priors closely. Francois highlights the importance of avoiding catastrophic forgetting in neural networks to progress towards AGI.