John Langford, a partner research manager at Microsoft Research New York, has left Microsoft, and he has declared the company's AI Frontiers group over. Langford announced the decision in a post on his blog, Machine Learning (Theory), which Digg summarized. He shared the post on social media on Nov. 6.
The group had about 50 people. It was moved out of Microsoft Research and into Microsoft AI in early September. Langford wrote that he found places for his team members as well as he could before leaving. His account does not detail the group's formal status or where the remaining staff were assigned.
What was Microsoft's AI Frontiers group working on?
Langford wrote that he and his colleagues created AI Frontiers inside Microsoft Research in 2023, seeking more organized research after the release of ChatGPT. Much of the group's work centered on making AI agents function reliably. He named Autogen, a framework for coordinating agents, and Fara, an agent that operates a computer to carry out tasks. Shunyatax also reported that some researchers in the group were exploring ways to improve models without simply making them larger.
Scaling versus research
Langford described the leadership philosophy in Microsoft AI as focused on scaling rather than research, according to Digg. He said the environment concentrated mainly on acquiring, creating and refining data.
He acknowledged the business logic behind that approach. In his assessment, returns to scale are more predictable than returns to research. Still, he called a scaling-only approach misdirected and unenjoyable. Those are his personal judgments, and they concern the unit his group joined, not company-wide policy. Shunyatax reported that Langford did not argue scaling should stop. Instead, he argued that scaling and fundamental research should run side by side.
No path to use
Why did John Langford leave Microsoft?
Langford wrote that he could not get his group's innovations used, or even seriously considered, while inside Microsoft AI. He said a return to Microsoft Research held little appeal, and he doubted research could reach practical use from there either.
He wrote that he is not comfortable working at a company where his research has no path to real use. He also said there will be no internships with the group in the coming year. He wished well to the colleagues he is leaving behind.
Langford's profile on the research platform alphaxiv lists him as a founding member of Microsoft Research New York. It also names him a co-creator of the Vowpal Wabbit machine-learning library. The Web Conference gave him a 2023 Test of Time Award for his contextual-bandit paper on personalized news recommendation. The paper describes algorithms that learn which article to show a reader.
Shunyatax placed the AI Frontiers news alongside other senior changes at Microsoft. It cited the planned departure of former LinkedIn chief executive Ryan Roslansky from his responsibilities over the Office and Teams divisions. It also cited the move of former AI platform head Eric Boyd to Anthropic.
Is the AI Frontiers exit linked to other Microsoft leadership changes?
The connection is unproven. The material does not establish that those changes stemmed from disagreements over research strategy, or that they are linked to Langford's exit.
Este artigo foi escrito com a assistência de IA.
News Factory APP - notícias agênticas para impulsionar seu SEO e AEO.
