Google’s Royal Hansen on AI: Where Are We? Where Are We Going?

Conversations with Bill Kristol

News, Society & Culture, Government, Politics

4.7 • 1.7K Ratings

🗓️ 26 July 2023

⏱️ 60 minutes

🔗️ Recording | iTunes | RSS

🧾️ Download transcript

Summary

With the recent rollout of ChatGPT and the development of other generative artificial intelligence, the future of AI—and its potential implications for national defense, economics, and society—has become a topic of urgent inquiry. To discuss where things stand with AI, we are joined by Royal Hansen, Vice President of Engineering for Privacy, Safety, and Security at Google. Hansen explains the breakthroughs achieved over the past decade, and the profound implications that powerful software and computing have brought to domains from language translation and music to medicine, national security, logistics, and other dynamic areas of the economy. Noting the potential of AI for use as well as misuse, Hansen explains why we need to develop thoughtful safeguards in the development of new technologies. According to Hansen, already we are benefiting from just the "low hanging fruit"—as seen in the developments in language translation and other generative AI available today. But the potential for breakthroughs in other fields likely are closer than might have been imagined just a few years ago.

Transcript

Click on a timestamp to play from that location

0:00.0	Hi. Welcome back to Conversations. I'm Bill Crystal. Very pleased to be joined again for the
0:19.6	second time in the last year by Royal Hanson, one of the most senior executives at Google. We discussed
0:27.6	privacy, safety, and security, which you're in charge of, I think, worldwide for Google. In December
0:33.3	of 2022, a conversation I recommend people look at and maybe be mostly reassured, I think, that
0:40.8	there, not everyone is, you know, the Gmail is safe, so to speak, and a million other things,
0:45.6	obviously, as well. But Royal has long experience in this whole area, though, of artificial
0:52.3	artificial intelligence and its precursors and what it means. And you've ever since you got your
0:56.8	degree in computer science at Yale, I guess, a little while ago. Yeah, I was thinking of David
1:01.6	Galertner and artificial intelligence in the early 90s, our old. And you studied with him, right?
1:05.6	Yeah. I studied with David, yeah. That's interesting. Yeah. So go back and look at that conversation
1:10.7	from December. I also, I discussed this with Jim Manzi, who works in this area almost exactly five
1:15.7	years ago, and I just skimmed the transcript of that last night. And I gotta say it stands up well
1:20.0	and raises questions, but also it's striking how things have changed since 2018. And Jim is a
1:24.4	pretty forward looking guy. It was certainly not minimizing what was going to happen. But
1:28.8	I think even he might be surprised by what's happened. So so chat GPT's been in the news,
1:32.7	artificial intelligence. Everyone's aware of these. I think that they've been breakthroughs or
1:38.1	changes, but what if where are we? What's what's happening here? How new is it? How surprising is it?
1:44.0	How fast is it moving? Et cetera. It's great. And I read that the transcript again too. And I thought
1:49.6	was really insightful and helpful to have a marker at that 2018 or so point because it is a paper
1:56.7	from Google in 2017 and the deep mind team that sort of sets us on the path we're on now. And I
2:04.2	think Jim captured that nicely. But even he, I suspect, couldn't quite have imagined
2:10.1	where this was all going. And I think there's always this balance between the hype of what's
	...

Please login to see the full transcript.

Previous episode | Next episode

Disclaimer: The podcast and artwork embedded on this page are from Conversations with Bill Kristol, and are the property of its owner and not affiliated with or endorsed by Tapesearch.

Generated transcripts are the property of Conversations with Bill Kristol and are distributed freely under the Fair Use doctrine. Transcripts generated by Tapesearch are not guaranteed to be accurate.