Exposing biases, moods, personalities, and abstract concepts hidden in large language models
MIT News • 2/19/2026
Summary
MIT has developed a new method aimed at exposing biases, moods, personalities, and abstract concepts hidden in large language models (LLMs). This method could help root out vulnerabilities and improve the safety and performance of LLMs.
Advertisement
Breaking Similar stories
Open Thread
Reason Magazine • 1 day ago
Global Cooperation in a Time of Geopolitical Disruption
Project Syndicate • Today
MIT faculty, alumni named 2026 Sloan Research Fellows
MIT News • Today
Mamdani Discovers Common Sense on Homeless Sweeps
City Journal • Today
For leftists, philanthropy is often simply an expression of plutocratic power. How then to make sense of the surprisingly radical Garland Fund?
Arts & Letters Daily • Today
“Rather than being a footnote to premodern folly, the Rosicrucian affair turns out to sit at the narrative center of the modern world”
Arts & Letters Daily • Today
DHS Spokesperson Tricia McLaughlin Goes Out at the Top of Her Game
Reason Magazine • Today
Babies Are Exposed to More Forever Chemicals in Utero Than Previously Thought
Nautilus • Today
The Haiku of Richard Wright
3 Quarks Daily • 1 day ago
Trump Administration May Grant Asylum to Turkish National Who Burned a Quran in the U.K.
Reason Magazine • Today
Survived Similar stories
Anti-Lindy Similar stories
Helping AI agents search to get the best results out of large language models
MIT News • 2 weeks ago
AI models were given four weeks of therapy: the results worried researchers
Nature News • 1+ months
Distinct AI Models Seem To Converge On How They Encode Reality
Quanta Magazine • 1+ months
A satellite language network in the brain
MIT News • 2 weeks ago
Using AI, Mathematicians Find Hidden Glitches in Fluid Equations
Quanta Magazine • 1+ months
These Bizarre, Centuries-Old Greenland Sharks May Have a Hidden Longevity Superpower
Scientific American Ideas • 1+ months
The Names of the Game: Moral Language and the Killing of Alex Pretti
3 Quarks Daily • 1 week ago
This Hidden Brain Region Could Help You Stay Resilient in Old Age
Nautilus • 1+ months
The Hidden Costs of L.A.’s “Mansion Tax”
City Journal • 1 week ago
A Hidden Lesson of the Minnesota Welfare Scandal
The Atlantic Ideas • 2 weeks ago