Jan Lecun believes OpenAI's Sora approach using generative models, focused on generating pixels, is ill-suited for simulating the world. He argues that the complexity of dealing with uncertainty in high-dimensional continuous sensory inputs makes generative models for sensory inputs likely to fail, contrasting their effectiveness in handling text data. Lecun and Meta propose their own approach, the video join embedding predictive architecture (VJPA), as an alternative solution to this problem.
Alongside Gemini 1.5's massive new context window, and Sora's mindblowing video generation, Groq has come along to redefine how fast we think LLMs can be. NLW explores people's reactions and the implications for new use cases.
INTERESTED IN THE AI EDUCATION BETA?
Learn more and sign up https://bit.ly/aibeta
Today's Sponsors:
Notion - Notion AI. Knowledge, answers, ideas. One click away. - https://notion.com/aibreakdown
ABOUT THE AI BREAKDOWN
The AI Breakdown helps you understand the most important news and discussions in AI.
Subscribe to The AI Breakdown newsletter: https://theaibreakdown.beehiiv.com/subscribe
Subscribe to The AI Breakdown on YouTube: https://www.youtube.com/@TheAIBreakdown
Join the community: bit.ly/aibreakdown
Learn more: http://breakdown.network/