Keynote: After the AI Hype – What’s Real, and What’s Next
summary
A clear timeline of what actually changed: neural nets win ImageNet in 2012 (from ~30–40% to 75% right away, 100% within two years), LLMs scale from GPT-2 (120M–1.5B) to GPT-3 (175B) to GPT-4 (~trillion), ChatGPT pulls in 100 million users in two months, Microsoft pours roughly $13B into OpenAI, and infrastructure and chip limits trigger overordering and a compute crunch. Concrete wins include AlphaFold (90% accuracy by 2020, broadly solved by 2022, 200M folds published) and early agentic coding tools that produce usable pull requests.
AI-written summary · about
chapters (13)
- 0:00AI confusion and history
- 3:49Science fiction shaping AI perception
- 8:02Modern generative AI pioneers
- 13:18Microsoft invests in OpenAI
- 18:34AI hype cycle explained
- 22:27Peak AI capabilities and multimodal
- 26:00GPT-5 limits and market consolidation
- 30:50Data center explosion effects
- 36:46Memory market collapse
- 39:58AI ethical risks and harm
- 43:13AI regulation and responsibility
- 45:41DeepMind protein folding
- 49:16Conclusion and AI future
related talks
43mBeyond the AI Hype: What's Real, What's Next
51mBeyond the AI Hype: What's Real, What's Next
52mBeyond the AI Hype: What's Real, What's Next
48mThe Perils and Promise of Artificial Intelligence
47mThe Rise of AI: A Reality Check on Generative AI
15mOne Year of Generative AI: What Has Happened and What's Next?
15mMore than 70 years of AI history in 15 minutes!
01h 01mKeynote: Transformers: The Rise of ChatGPT