Watch now | Chatted with John Schulman (cofounded OpenAI and led ChatGPT creation) on: How post-training tames the shoggoth & enabled GPT-4o AI coworkers in 1-2 years The plan if AGI comes in 2025 Reasoning traces, long horizon RL, & multimodal agents Plateaus & moats
Thanks Dwarkesh really appreciate this. Although I am beginning to feel a real "law of diminishing returns" on the topic of when will AGI arrive and what will we do so it would be great if future podcasts didn't dwell too much on this area. I don't think anyone can give a thorough answer at this stage so it feels like we just need to "feel our way towards it".
I enjoyed the bits when you guys talked about models, pre and post training. Less from a "moat" perspective and more in terms of what is possible / what will change / how these approaches will develop. Keep up the good work
Thanks Dwarkesh really appreciate this. Although I am beginning to feel a real "law of diminishing returns" on the topic of when will AGI arrive and what will we do so it would be great if future podcasts didn't dwell too much on this area. I don't think anyone can give a thorough answer at this stage so it feels like we just need to "feel our way towards it".
I enjoyed the bits when you guys talked about models, pre and post training. Less from a "moat" perspective and more in terms of what is possible / what will change / how these approaches will develop. Keep up the good work
I enjoyed the interview, good stuff 👍