news
MIT News
November 26th, 2025 at 5:00 AMResearchers discover a shortcoming that makes LLMs less reliable
Large language models can learn to mistakenly link certain sentence patterns with specific topics — and may then repeat these patterns instead of reasoning.
#ai
#llm
#research
Get stories like this every Monday
The week's most important AI news in one email. Free, unsubscribe anytime.
Score: 1.98
Engagement proxy: 0
Canonical link: https://news.mit.edu/2025/shortcoming-makes-llms-less-reliable-1126