Robert discusses the challenges of intrinsic motivation in deep learning models, highlighting the issue of agents getting stuck in unchanging environments. Keith points out the importance of information transmission among fish in goal-rewarding states, suggesting a need for further exploration in this area.
