Why Sustainable? AI And Our Shared Challenges

If this headline got you to think, read on.

We are in the midst of trying to figure out how to use AI – in business, in education, in defense (of a nation or the whole great tribe of humanity) of science … and how do we answer the questions that bedevil those who are thinking years in the future?

I saw a segment at the Next Endeavor event in Mountain View, California, which Imagination in Action and partners put on (disclaimer: I help to run these events) this September, where two Meta people, Lisa Rivalin and Richa Mishra, talked about how the company is making moves forward in this area.

The Meaning of Sustainability

Rivalin started by describing the duality behind the term “sustainability,” and another duality in how we use AI for that objective set.

“There’s a double meaning of ‘sustainable’ here, meaning, something that lasts long, and also reducing the impact on the environment,” she said. “(Then there’s) using AI to solve environmental problems, and … sustainable AI, which is reducing the environmental footprint of the AI hardware itself.”

Mishra evoked a metaphor:

Users, she noted, shouldn’t see all of this.

“As infrastructure engineers, our biggest success is being invisible,” she said. “And the only time you realize we’re there is when something fails.”

As for energy, Rivalin suggested that some of these problems are popping up anew.

“It’s more visible now because of the accelerating demand,” she said. “So there is an elephant in the room. If we need to build more data centers, we have more needs in terms of hardware.”

Rivalin mentioned the role of supply constraint.

“It’s really, really hard to actually get semiconductors right now,” she said.

Of Progress Over Time

How It Works

“A lot of agents are doing the work for humans, and humans are making decisions,” Rivalin continued. “But before we actually had to reach this state, we had to go through a set of principles.”

“Create smaller agents.”

People and More

She had this to say about the grid:

Rivalin hyped collaboration: