NYT reporter Cade Metz Q&A on new book on AI, pioneers

As a kid growing up in Raleigh, Cade Metz was surrounded by people who were making huge technological advances but remained nearly anonymous to wide swaths of the public.

His parents both worked at IBM’s campus in Research Triangle Park, where patents and new inventions were constantly being spun out.

His father worked closely with people like George Laurer, the Raleigh-based engineer who created the universal bar code in the early 1970s.

Not many people knew about the George Laurers of the world, even though they were revolutionizing the way society went about their lives, including shopping. That’s something Metz, now a technology reporter for The New York Times, always has tried to change with his reporting.

“What I’ve always wanted to do was not necessarily write about the entrepreneurs who people so often write about when it comes to tech industry,” Metz said in a telephone interview with The News & Observer. “You know, the founders and the big personalities, like Steve Jobs. But rather the people actually building the technology. I think they are underrepresented.”

With his new book, “Genius Makers: The Mavericks Who Brought AI to Google, Facebook, and the World,” Metz is sharing the stories of the individuals whose breakthroughs in the past decades have pushed forward the limits of artificial intelligence.

As a reporter for Wired Magazine and now the Times, Metz has had a front row seat to the incredible strides made in deep learning, a process in which a computer learns from absorbing large amounts of data. That has helped computer programs make huge leaps in things like facial recognition and self-driving cars, and fueled intense competition between companies, like Google, Microsoft, Facebook and multiple Chinese firms.

“One of the things that I want to do with the book is give people a really good idea of what has actually happened” with AI, he said. “And where we’re going.”

In a different world, Metz, a 48-year-old graduate of Broughton High School and Duke University, would be traveling back to his hometown to kick off the launch of his new book.

But instead, he will be hosting multiple virtual events with Raleigh companies.

On March 16, also the publishing date for the book, Quail Ridge Books will host a virtual conversation at 7 p.m. between Metz and UNC professor and novelist Daniel Wallace.

The News & Observer spoke with Metz to talk about his book.

Editor’s Note: This conversation has been edited and condensed for space and clarity.

Q. How did your parents’ jobs at IBM influence your career in journalism?

Metz: What I tell everybody, is that I’m very much a product of both my parents. My mother was actually an economics major, but she very much had an English major sensibility. She had a bookshelf in our living room filled with first editions. So, from an early age, I wanted to be a writer and I wanted to study English, which I did at Duke.

But I also had a scholarship through IBM. And that meant that I interned at IBM during the summers as a programmer, where I was testing network cards, and writing C programs. And, you know I honestly feel like the book is a culmination of that.

Q. What characteristics stand out to you about the people you profile in your book?

Metz: Many of these people are very idealistic. They saw their technology — the technology they nurtured over the decades — as something that should not be used for particular purposes. And then, you know, they were sucked into industry, and they were forced to confront all that.

There was a big controversy at Google over the use of these ideas on a (Department of Defense) project called Project Maven, which was used to identify objects like vehicles and buildings and people from drone footage. There was all this concern at Google among certain employees that this would be used in the battlefield. So there was a big protest and Google ended up pulling out of it.

The book ends up being about issues like that, where you’ve got sort of the idealism of these developers that sometimes clashes with the aims of these giant companies, which are so, so wealthy and so powerful and have their own aims — a lot of driven by money.

Q. What else are these AI creators grappling with?

Metz: The other big area, which is very much in the news now, is the issue of bias in these systems. And what you see, and have seen over the years, is that a lot of the researchers in this field were white men. That created its own complications in that these neural networks (behind AI) literally train on data to learn their tasks, whether it’s image recognition or speech recognition or natural language understanding.

So if you want to build a system that can recognize a cat, you feed that neural network thousands of cat photos and it learns to recognize the patterns that define a cat. Well, what that means is that the people building the systems are choosing the data that’s going to define the system. And if they’re all white men, for instance, there’s a certain point of view, and that can affect the system.

What we’ve seen in recent years is that some of these systems, like facial recognition systems, can be biased against women and people of color or chat bots that use racist language. This is what I want to convey to people is that a neural network learns from a human-generated data and what that means is it learns from our deficiencies, as well as our strengths.

That’s an issue, like how to weed out all that bias, that hasn’t been resolved.

Q. You’ve reported on Google firing two AI researchers who had criticized the company’s approach to examining bias in its AI. Do you feel like we are reaching an inflection point in how we think about AI?

Metz: It’s definitely an inflection point, and the place to see this is at Google, where two AI ethics researchers were recently ousted from the company, Timnit Gebru and Meg Mitchell. They were at the forefront of calling attention to this problem. And, you know, for various reasons, their superiors at Google didn’t see eye to eye with them, and they’re now gone from the company.

That’s a big issue at Google, but it’s also an issue across the industry. And a lot of companies are going to have to contend with this, and this is coming to a head right now. And my book, in part, is about that. There’s a whole chapter where Timnit and Meg are characters, and they are confronting these companies with these issues. I think you’re seeing people, on both sides of the issue, really start to speak up and really clash over this particular problem.

Q. Do you think most people understand how far AI has come in the past decade?

Metz: AI gets thrown around so much, and it gets applied to almost every technology there is. And what that means is that people don’t have a good understanding of what’s really going on.

What we’ve seen over the past 15 years is the rise of neural networks that help with speech recognition and image recognition. And that’s been applied in really impressive ways, from self-driving cars and other robotics to natural language understanding chat bots that are coming to the fore that can generate their own language.

But when you walk out your front door, you don’t see self-driving cars everywhere. The progress of that was, to say the least, exaggerated. That problem is so much harder than people realize, and that is nowhere near a system that can do anything the human brain can do. Even though we’ve had all this progress, there’s still so much that needs to be solved in these particular areas, and that includes self-driving cars and that includes chat bots.

We’re certainly nowhere close to a system that that can do anything that the human brain can do, but there are two labs in the book — OpenAI and DeepMind — whose stated mission is to do that. They’re doing incredible work, but they don’t quite know when that might happen. That could be decades, or even centuries away. It’s hard to predict.

Q. Do you think there’s more fear around AI than there should be?

Metz: Everything we’re discussing now — this idea of machines that can do anything a human brain can do and the worry over what that might be — are the same discussions that have gone on since the ’50s, when the term AI was coined.

There’s this repeated over-optimism in the field. It’s just part of it, and what ends up happening is that optimism swells and recedes when people finally realize that technology isn’t as advanced as everyone thought it would be.

They call that an “AI winter,” and the book kind of shows that over the years, these ideas ebb and flow. Right now the progress is continuing, but it’s really hard to extrapolate into the future. Naturally, none of us can really know what’s going to happen in the future. What I will say is that the path to some of the stuff that is talked about is still unclear, even with things like self-driving cars.

Q: How often do you make it back to the Triangle?

Metz: I haven’t over the past year for obvious reasons. But normally we get back a lot. My wife is from Raleigh. My mother’s still there living in Cameron Park, where I grew up. We always go to the beach, too, down at Ocean Isle. We spend a lot of time here, if only because we miss the North Carolina barbecue cause you can’t get it anywhere else in the world.

Q: Which one is your favorite?

Metz: When I was growing up, the ultimate was always Wilber’s (in Goldsboro). And my wife’s family’s favorite was always Kings in Kinston. But then, whenever I am in Raleigh, I love Clyde Cooper’s.

Virtual book events

▪ On March 16, also the publishing date for the book, Quail Ridge Books will host a virtual conversation at 7 p.m. between Metz and UNC professor and novelist Daniel Wallace. Go to quailridgebooks.com for ticket info.

This story was produced with financial support from a coalition of partners led by Innovate Raleigh as part of an independent journalism fellowship program. The N&O maintains full editorial control of the work. Learn more; go to bit.ly/newsinnovate