Artificial Intelligence

OpenAI, DeepSeek, and Google vary widely in identifying hate speech

Neil Fasching and Yphtach Lelkes of the Annenberg School for Communication have found dramatic differences in how large language models classify hate speech, with especially large variations for language about certain demographic groups, raising concerns about bias and disproportionate harm.

2 min. read

AI goes to college: How new tech is driving majors and jobs

Chris Callison-Burch of the School of Engineering and Applied Science looks at the explosion of interest in AI from undergraduates.

AI’s growth potential

Alexander Arnon of the Wharton School and colleagues predict that generative A.I. will lift productivity and economic growth during the next decade, followed by trailing returns.

AI is supercharging Gen Z workers — if they can land a job

Ethan Mollick of the Wharton School says that young workers might miss the chance to learn how senior colleagues think through problems by instead asking an AI chatbot.

Ethan Yang, Jjianjing Kuang, Kevin Li, and Henry Huang.

Where AI models fall short in mimicking the expressiveness of human speech

Undergraduates Kevin Li, Henry Huang, and Ethan Yang worked with linguistics professor Jianjing Kuang to compare speech production and perception by humans and AI—research that can help companies move closer to natural and expressive AI speech.

2 min. read

How to make AI do bad stuff: Treat it like a person and sweet talk it

Researchers from the Wharton School and School of Arts & Sciences have discovered how to manipulate powerful AI models with the same psychological tricks that work on humans.

A screenshot of a Replika chatbot screen.

What is real about human-AI relationships?

In a new paper, Annenberg School for Communication doctoral student Arelí Rocha explores how people discuss their relationships with AI chatbots.

2 min. read

AI uncovers new antibiotics in ancient microbes

César de la Fuente uses AI to hunt for new antibiotic candidates in unlikely places, from the DNA of extinct organisms to the proteins of ancient microbes.

2 min. read

Chinmay Govind writing equations on a chalkboard.

Students use machine learning to track and protect whale populations

For their Penn Undergraduate Research Mentoring Program project, Chinmay Govind and Nihar Ballamudi leveraged AI to locate and census whales. The study may inform policy measures that help to improve protections for whale populations worldwide.

5 min. read

An MIT report that 95% of AI pilots fail spooked investors. But it’s the reason why those pilots failed that should make the C-suite anxious

Ethan Mollick of the Wharton School is referenced for his theory that the benefits of AI will come when companies abandon getting AI models to reflect existing processes.

Filters

OpenAI, DeepSeek, and Google vary widely in identifying hate speech

AI goes to college: How new tech is driving majors and jobs

AI’s growth potential

AI is supercharging Gen Z workers — if they can land a job

Where AI models fall short in mimicking the expressiveness of human speech

How to make AI do bad stuff: Treat it like a person and sweet talk it

What is real about human-AI relationships?

AI uncovers new antibiotics in ancient microbes

Students use machine learning to track and protect whale populations

An MIT report that 95% of AI pilots fail spooked investors. But it’s the reason why those pilots failed that should make the C-suite anxious