Anthropic co-founder Jared Kaplan on scaling and the road to human-level AI. At AI Startup School in San Francisco. 00:17 - From Physics to AI 01:41 - Initial Skepticism and Shift to AI 02:12 - AI Training Phases 02:32 - Pre-Training 03:16 - Reinforcement Learning 04:19 - Scaling Laws in Training 08:19 - Unlocking AI Capabilities 11:27 - Organizational Knowledge and Memory 12:19 - Oversight and Nuanced Tasks 13:38 - Preparing for the Future 15:48 - Claude 4 and Beyond 21:18 - Human-AI Collaboration 29:50 - Scaling Laws and Compute Efficiency 35:26 - Audience Q&A
@ycombinator Physics is different, AI is also different. But it's possible to blend them both. Scaling is something to look forward to. However, there'll definitely be hookups, that's almost a certainty.
@ycombinator Intriguing... Pre-training alone won't cut it though, real-world deployment adds another layer of complexity + refinement.. makes u think 🤔
@ycombinator Physics to AI is a natural transition - both fields seek fundamental patterns in complex systems. Excited to hear his insights on scaling.
@ycombinator Fascinating how Kaplan frames AI development as phases rather than breakthroughs. Makes me wonder if we're still in pre-training for AGI.
@ycombinator Fascinating insights from Jared Kaplan on how scaling laws are propelling AI toward transformative capabilities. It's a reminder that persistence in research can turn skepticism into breakthroughs excited to see where this leads next!
@ycombinator @grok summarise this video for me. identify key topics and provide me structured summary of each topic. if one tweet is not enough then extend summary to multiple tweets.
@ycombinator Scaling laws will push AI forward, but Canada's R&D stagnation guarantees talent drain. We're not building the infrastructure to compete in this race.
@ycombinator Scaling breakthroughs make real-time AI applications so exciting for biz.
