Press

Anthropic is hiring a smaller share of junior technical staff, while AI agents' working hours now outnumber its employees' six to one, according to a report by professor Nathan Wilmers and co-authors. "The share of junior-level employees in technical departments at Anthropic slipped from 21% to 14% since early 2025," the report stated. The findings were based on internal data and employee interviews.

"AI has made it more important for the CIO to be able to exercise influence and give advice," said principal research scientist George Westerman. "Now you have a major technology that every C-suite person is wondering about and feels they need to use, and, given its transformative nature throughout the organization, this is where the skills of a CIO who is a strategic partner can be really useful."

Corporate labs' retreat led basic research to become much more reliant on government-backed efforts. In the view of private-science advocates, that is lamentable. A study of 100,000 National Institutes of Health (NIH) grant applications by Pierre Azoulay and co-author found that projects that took out-of-consensus approaches were less likely to have their funding renewed. 

A servant leader mindset — coupled with the results she has helped produce — are among the reasons why assistant dean Dawna Levenson is the recipient of P&Q's 2026 Best in Class Award for MBA Admissions. Levenson represents the spirit of the program: collaborative, creative, practical — and never afraid to roll up her sleeves to get things done. More than that, this approach rubs off on everyone who comes in contact with her. As a whole, Levenson feels privileged to be the assistant dean for admissions at MIT Sloan. That's because she gets to see the "positive outcome of the work her team has done every single day."

A study by professor Jackson G. Lu and co-authors found that OpenAI and China's Baidu models responded differently to the same prompts depending on whether they were written in English or Chinese. The language of a prompt influenced the cultural assumptions underlying its response. "I doubt that we could remove every cultural tendency from a model, because language and culture are deeply intertwined," said Lu.

A study by professor Jackson G. Lu and co-authors found that OpenAI and China's Baidu models responded differently to the same prompts depending on whether they were written in English or Chinese. The language of a prompt influenced the cultural assumptions underlying its response. "I doubt that we could remove every cultural tendency from a model, because language and culture are deeply intertwined," said Lu.