tech
AI professors are negotiating the new realities of academic research
At a convening for the Schmidt Sciences AI2050 program, I saw how academic researchers are facing up to the challenges of the AI era.
TL;DR
- The cutting edge of AI research has shifted from universities to private companies that can afford the extensive resources needed for large language models (LLMs).
- Academic researchers struggle to access and study proprietary frontier models like ChatGPT and Claude, impacting their ability to conduct detailed research.
- Many university AI researchers are redirecting their efforts to niche problems that industry may overlook, such as ethical considerations or specialized AI models for specific scientific domains.
- The cost of accessing models and limited federal funding are significant concerns for academic researchers.
- There is a growing concern that AI may automate mathematics, impacting the future of pure math research.
- Some researchers believe that AI could enhance human scientists' efficiency and that resource constraints may foster innovation in model efficiency and architecture.