Hyundai Motor Group and Boston Dynamics recently announced the launch of the Boston Dynamics AI Institute, with more than US $400 million to start. Boston Dynamics founder Marc Raibert is to head it up, for the purpose of “creating future generations of advanced robots and the world’s smartest and most capable intelligent machines.”

Boston Dynamics, founded in 1992, is already famous for its robot dogs:

Raibert has bigger plans for smarter robots, as he detailed in a recent interview, as he told Evan Ackerman at IEEE Spectum:

The new thing that’s clearly different from what Boston Dynamics is doing, is to make robots smarter, in the sense that they need to be able to look at the world around them and fundamentally understand what they’re seeing and what’s going on. Don’t get me wrong. This is currently science fiction, but I’ve learned that if you keep working on something like this long enough with enough resources, you may be able to make progress. So, I’d like to make a robot that you can take into a factory, where it watches a person doing a job, and figures out how to do that job itself. Right now, it takes a fleet of programmers even for simple tasks, and every new thing you want your robot to do is a lot of work. This has been clear for years, and I want to find a way to get past that. And I don’t think the lay public understands how stupid robots are compared to people—a person could come into my workshop and I could show them how to do almost any task, and within 15 minutes, they’d be doing it. Robots just aren’t anything like that…yet.

Evan Ackerman, “Boston Dynamics’s founder thinks creatively with $400 million AI Institute” at IEEE Spectrum (August 17, 2022)

The really interesting question is, is that just a matter of developing new technology? The human observer has an underlying psychology, which includes his own motives and goals and need to fit into a social structure he cares about. Can that be duplicated in a non-living, non-thinking entity? We shall see.

Raibert also wants his new institute to tackle ethics issues:

There are four topics that always come to mind for me. One is the use of robots by the military, one is robots taking jobs, and one is killer robots (or robots that are intended to harm people without human-in-the-loop regulation), and one is the idea that robots will somehow take over the world against the will of human beings. I think the last two are where you get the least grounding in what’s really happening, and the others are works in progress. The military topic is a very complex thing, and with the jobs topic, yes, some people’s jobs will be done by robots. Other jobs that don’t yet exist will be created by robots. And robots will help people’s existing jobs become safer and easier. I hope we’re going to be open about all of these things—I’m not embarrassed about my opinions, and I think if we can have an open conversation, it’ll be good.

Evan Ackerman, “Boston Dynamics’s founder thinks creatively with $400 million AI Institute” at IEEE Spectrum (August 17, 2022)

Military robots, as Robert J. Marks pointed out in The Case for Killer Robots, are likely to happen no matter what. Raibert is likely right in thinking that robots will also create many jobs — eliminating the drudgery frees people’s time for more creative and life-enhancing pursuits, leading to higher living standards.

That said, the use of robots for surveillance and policing, especially in unfree nations, could soon be an ethical issue, especially for those who manufacture them:

A robot need not be very smart to be seen as a menace:

At any rate, there won’t likely be a shortage of ethical issues for the new institute to ponder.

You may also wish to read:

Boston Dynamics’ famous robot dog being put to work. Long stalled in the area of research and development, “Spot” is now being prepared for its first job. It’s easy to imagine robots like this monitoring a large fraction of our infrastructure in the future, but some are using the technology to build weapons. (Jonathan Bartlett)

and

Turns out, computers are not vacuuming up all our jobs… Far from it, we can hardly find all the people we need to manage the computers. Jeffrey Funk and Gary Smith argue that the best results come from humans using AI — it certainly worked for radiologists.