Featured Post from University of Rochester

Why generative AI 'hallucinates' and makes up stuff

University of Rochester’s Christopher Kanan says current iterations of AI lack human-like self-awareness and reasoning abilities.

Apr 10, 2025

2 min

Christopher Kanan

Generative artificial intelligence tools, like OpenAI’s GPT-4, are sometimes full of bunk.


Yes, they excel at tasks involving human language, like translating, writing essays, and acting as a personalized writing tutor. They even ace standardized tests. And they’re rapidly improving.


But they also “hallucinate,” which is the term scientists use to describe when AI tools produce information that sounds plausible but is incorrect. Worse, they do so with such confidence that their errors are sometimes difficult to spot.


CaptionResizeWrap TextRemove

Christopher Kanan, an associate professor of computer science with an appointment at the Goergen Institute for Data Science and Artificial Intelligence at the University of Rochester, explains that the reasoning and planning capabilities of AI tools are still limited compared with those of humans, who excel at continual learning.


“They don’t continually learn from experience,” Kanan says of AI tools. “Their knowledge is effectively frozen after training, meaning they lack awareness of recent developments or ongoing changes in the world.”


Current generative AI systems also lack what’s known as metacognition.


“That means they typically don’t know what they don’t know, and they rarely ask clarifying questions when faced with uncertainty or ambiguous prompts,” Kanan says. “This absence of self-awareness limits their effectiveness in real-world interactions.”


Kanan is an expert in artificial intelligence, continual learning, and brain-inspired algorithms who welcomes inquiries from journalists and knowledge seekers. He recently shared his thoughts on AI with WAMC Northeast Public Radio and with the University of Rochester News Center.


Reach out to Kanan by clicking on his profile.


Connect with:
Christopher Kanan

Christopher Kanan

Associate Professor of Computer Science

Christopher Kanan's research focuses on deep learning and Artificial Intelligence (AI)

AI and Machine LearningApplied Machine Learning (e.g. Medical Computer Vision)Language-guided Scene UnderstandingArtificial IntelligenceDeep Learning

You might also like...

Check out some other posts from University of Rochester

2 min

Adam Frank: New Peer-reviewed Studies Change the Conversation on UFOs

For decades, talk of UFOs has thrived on fuzzy photos and personal anecdotes—never the kind of hard data scientists can actually test. But new peer-reviewed studies have changed the conversation, says Adam Frank, a University of Rochester astrophysicist who studies life in the universe and the nature of scientific discovery. Two recent papers, published in reputable astronomy journals, claim to have found evidence of “non-terrestrial artifacts” in astronomical photographs from the 1950s — objects that appear to be  orbiting Earth before the Space Age began. “That’s an extraordinary claim,” Frank says, “and, as Carl Sagan famously said, 'Extraordinary claims require extraordinary proof.' “The good news is that, finally, there’s something associated with UFOs that science can work with.” Led by astronomer Beatriz Villarroel and her VASCO project (Vanishing and Appearing Sources during a Century of Observations), the studies passed the first test of scientific credibility: rigorous peer review. Now, Frank says, comes the harder part — the “call-and-response” that defines real science. “Getting a paper published doesn’t make the claim right,” he explains. “It just means the debate can begin. Other scientists will now dig into the data, test the methods, and try to tear the claim apart. That’s how science works.” Frank is a frequent on-air commentator for live interviews and segments in national media outlets and the author of The Little Book of Aliens (Harper Collins, 2023). He also regularly contributes to written publications, including The Washington Post, The Atlantic, The New York Times, and Scientific American. In 2021 he received the Carl Sagan Medal, which recognizes and honors outstanding communication by an active planetary scientist to the general public. It is awarded to scientists whose efforts have significantly contributed to a public understanding of, and enthusiasm for, planetary science. Connect with him by clicking on his profile. 

1 min

'Brain-on-a-chip': Engineering tomorrow’s breakthroughs today

A “brain-on-a-chip” technology might sound like science fiction, but it’s real-world hope. James McGrath, a biomedical engineer at the University of Rochester, leads a team that develops micro-scale tissue chips to study diseases in lieu of conducting animal experiments. The team’s “brain-on-a-chip” model replicates the blood-brain barrier — the critical membrane separating the brain from the bloodstream — to mimic how the barrier functions under healthy conditions and the duress of infections, toxins, and immune responses that can weaken it. Recent findings from McGrath’s team show how systemic inflammation, such as that caused by sepsis, can compromise the barrier and harm brain cells. The researchers also demonstrated how pericytes — supportive vascular cells — can help repair barrier damage, an insight that could guide new therapies for Alzheimer’s and Parkinson’s. The research culminated in a pair of recent studies published in Advanced Science and Materials Today Bio. “We hope that by building these tissue models in chip format, we can arrange many brain models in a high-density array to screen candidates for neuroprotective drugs and develop brain models with diverse genetic backgrounds,” McGrath says. McGrath aims to transform how scientists test drugs and predict neurological side effects before they occur — helping rewrite how we study, and one day safeguard, the brain. Contact McGrath by clicking on his profile

1 min

Government Shutdown: With Senate in the spotlight, expert Gerald Gamm offers insight

The Senate returned to Capitol Hill on the first day of a government shutdown to vote on two funding bills aimed at getting the government up and running amid an ongoing blame game among congressional leaders. University of Rochester political scientist Gerald Gamm is watching the deliberations and political maneuverings closely and is in a unique position to lend insight on the negotiations and gamesmanship. Gamm is a co-author of Steering the Senate (Cambridge University Press, June 2025). The book has received high praise from a multitude of sources, and has been called "essential reading for all who care — or worry — about the past and future of institutional leadership and capacity on Capitol Hill," "the best book we have about the organizational development of the Senate," and "a masterpiece . . . that unearths new information on the emergence of leadership institutions and the role of parties and showing their relevance for the Senate of today." Gamm is available for interviews and can be contacted by clicking on his profile.

View all posts
Powered by