Research conducted by Stanford University has uncovered a concerning trend in the field of artificial intelligence: the rapid development of large language models (LLMs) is significantly outpacing the research aimed at understanding them. The study evaluated 50 widely cited claims about LLMs against 20 current models and found that 44 percent of these findings have already become outdated, with another 34 percent showing considerable weakening in relevance.
The average lifespan of claims about LLMs is now just 15.7 months, with assertions regarding the limitations of these models expiring the fastest. This stark reality poses challenges for researchers and practitioners who rely on these claims for guidance in their work. As AI technology continues to advance at an unprecedented pace, the need for ongoing research to keep pace with these developments becomes increasingly critical.
OpenAI’s recent hiring of Dan Driscoll, a former Army Secretary, suggests a strategic move to enhance bipartisan engagement and influence in shaping AI regulations, particularly within military and intelligence contexts. This hiring aligns with the pressing need for informed governance as AI technologies evolve rapidly.
Compiled automatically by the Tech AI Newsdesk from public AI-news sources and summarised in our own words.
