Microsoft debuts supercomputer for creating human-like artificial intelligence

Microsoft on Tuesday announced that it has built one of the top five supercomputers in the world designed specifically for OpenAI. Yahoo Finance’s Dan Howley joins Seana Smith to discuss.

Video Transcript

SEANA SMITH: We want to get the news out of Microsoft today, the company announcing that it has built one of the top five supercomputers in the world. Dan Howley is looking into this for us. And Dan, I guess what can we expect this computer to be able to do?

DAN HOWLEY: Yeah, so they announced two big things at their Build Conference. Usually this is held live in Seattle, but obviously given the coronavirus, it's being held virtually this year. So Satya Nadella and company basically announced one of these new supercomputers running on Azure, which is its cloud platform.

And it's built for specifically the OpenAI Group, and that's essentially a group that looks to build general artificial intelligence. That's artificial intelligence that basically would think like a person. It's the kind of AI that you hear about in disaster movies, almost. But OpenAI Initiative's goal is to ensure safety, that the innovations that they put into this aren't done without roadblocks in the way, or aren't done without thinking through what the issues might be that providing this kind of intelligence to machines could do. So the supercomputer is meant to enable that kind of research going forward.

And look, it's nowhere near reaching human levels yet. They have it doing things like creating music, or-- the music doesn't sound great, by the way-- or solving Rubik's Cubes. But it is on its way to getting smarter. And that's something that I think Microsoft is smart to invest in.

And on the other side, it's also doing some investing in anti-bias tools for artificial intelligence, basically machine-learning algorithms. We've seen these kind of biases kind of sneak into machine learning. What happens is the developers have their own ingrained biases, whether purposeful or not. Most of time it's not. It's simply just a matter of not thinking through all the different variables that can result in an outcome.

For instance, they've worked with EY. And they found that men were being, in certain scenarios, in their test cases, were being awarded loans more often than women. So they used this kind of anti-bias tool, and managed to knock it down to less than a percent of the time men were being awarded more than women, these loans.

And again, this was just a test. It wasn't real loans going out there. So nothing to worry about there. But it is just proof of concept that this kind of tool could help take bias out of these machine-learning algorithms going forward.

SEANA SMITH: Yeah, two interesting announcements from Microsoft today. Dan Howley, thanks so much for breaking that down.