Two AI pioneers, Ilya Sutskever and Yann LeCun, argue that simply scaling large language models with more GPUs is reaching its limits. Sutskever believes the industry is transitioning from an "age of scaling" to an "age of research," where new ideas matter more than raw compute power. LeCun goes further, claiming LLMs aren't
Table of contents
1. Sutskever’s Timeline: From Research → Scaling → Research Again2. Why the Current LLM Recipe Is Hitting Limits3. Safe Superintelligence Inc.: Betting on New Recipes4. Have Tech Companies Overspent on GPUs?5. Yann LeCun’s Counterpoint: LLMs Aren’t the Future at All6. Sutskever vs. LeCun: Same Diagnosis, Different Cure7. What All This Means for Developers and Founders8. A Quiet Turning PointSort: