Stanford, California, USA - March 17, 2019: Aerial view of Stanford University in Stanford California. Stanford is a private university founded in 1885 by Leland and Jane Stanford.

getty

News out of Harvard shows how researchers there are joining forces to zero in on AI that is powerful, safe and equitable.

The Stanford Data Science initiative, a Stanford program focused on data science research and education, and the Stanford Institute for Human-Centered AI, which, as its name implies, focuses on human impact, will now be run under one roof, with the Stanford HAI moniker attached.

HAI brings together researchers from computer science, medicine, law, education, business, and the humanities. They’ll work on technical AI advances alongside ethics, policy, safety, and societal impact. The data science program focuses on collaboration, education, and research involving statistics, computing, and large-scale data analysis in fields ranging from medicine and biology to engineering and social science. Both of these are important. The current roster shake-up is intended to support that reality.

Combined Research Power

Stanford HAI has some of the best scientists around, and lots of grant money. The Stanford Data Science initiative has the Marlowe computing cluster, an NVIDIA DGX H100 SuperPOD with 248 H100 GPUs and storage measured in petabytes.

In a press statement, Harvard President Jonathan Levin called the new combined office the “front door for AI at Stanford.”

“The merged organization creates a community of scholars whose research touches powerfully on every aspect of AI, its applications, and implications, and the human-centered focus provides a north star for the institute,” Levin said.

Big Names

In terms of specific leadership, one name that stands out is James Landay, who will head the newly centralized lab. Landay’s decades of work on human-computer interaction is well-regarded. Then there’s Fei-Fei Li, who is known for contributions such as pioneering modern computer vision through the creation of ImageNet, and had a leadership role at Stanford HAI. Li will be a co-chair of the institute’s advisory council, along with John Hennessy, and will also work as Levin’s Special Advisor on AI.

Now, this little tidbit caught my attention, as I concede I’m less familiar with Stanford’s internal community than the one here at MIT. Landay is listed as the “Denning director,” which for a second I thought was some kind of specialized title.

It turns out that Steven A. Denning was a major Stanford donor, venture capitalist, and former chairman of Stanford’s Board of Trustees. That’s where that part comes from.

As for Landay, he had this to say about the consolidation:

“This technology is changing everything. To have real impact in this moment, we need to adapt. This is about shaping how AI affects people, communities, and society – with that human-centered perspective at the core of everything we do.”

A Multi-Disciplinary Hub

I’m struggling to tell this part of the story in color.

At the new Stanford HAI, participants will be using Marlowe and other tools to drive progress in many different types of projects: to improve self-driving vehicle technology, for example, or to bring history to life with new curricula, or to find a better way to serve rural populations in different service realms. The list goes on.

Here’s how it’s described in the relevant Stanford Report online:

“Neuroscientists build models that predict brain activity. Historians run natural language processing over archival collections to surface patterns in how societies communicate. And education researchers test tutoring systems that adapt to individual learners and support teachers in classrooms.”

That’s a lot – all powered by technologies that, as lately as the 2000s, sounded like science fiction. We’re really scrambling to make room for AI at the table, and the Stanford program is just one of many examples, happening, in unison, right now.