News Archive - Ioana Patringenaru

A New Method to Steer AI Output Uncovers Vulnerabilities and Potential Improvements

A team of researchers has found a way to steer the output of large language models by manipulating specific concepts inside these models. The new method could lead to more reliable, more efficient, and less computationally expensive training of LLMs. But it also exposes potential vulnerabilities.

How Much Does Chatbot Bias Influence Users? A Lot, It Turns Out

Customers are 32% more likely to buy a product after reading a review summary generated by a chatbot than after reading the original review written by a human. That’s because large language models introduce bias, in this case a positive framing, in summaries. That, in turn, affects users’ behavior.

After a Heart Attack, the Heart Talks to the Brain - and the Brain Responds

Researchers at the University of California San Diego have discovered an unexpected communication circuit between the heart, brain and immune system that shapes how the heart responds after a heart attack.

First-of-its-kind Cleanroom Turns Inventions into Devices Ready for FDA Approval

A first-of-its-kind good manufacturing practices facility located on a university campus in the United States opened this fall at the University of California San Diego. The space is dedicated to building devices that can be implanted in the human body – especially for neurological applications.

Take a Tour of UC San Diego’s New Good Manufacturing Practices Facility

The new good manufacturing practices facility transforms UC San Diego neurotechnology discoveries into implants for the brain, spine and eye that meet FDA manufacturing standards

The Sky is Full of Secrets: Glaring Vulnerabilities Discovered in Satellite Communications

With $800 of off-the-shelf equipment and months worth of patience, a team of U.S. computer scientists set out to find out how well geostationary satellite communications are encrypted. And what they found was shocking.

From Chatbots to Dice Rolls: Researchers Use D&D to Test AI’s Long-term Decision-making Abilities

Large Language Models, like ChatGPT, are learning to play Dungeons & Dragons. The reason? Simulating and playing the popular tabletop role-playing game provides a good testing ground for AI agents that need to function independently for long stretches of time.