00:00I was pretty surprised by these findings. I would assume that most people knew that the U.S. was ahead.
00:04Were you similarly surprised? And what do you think is driving this mismatch in perception?
00:08I was surprised. I think I was most surprised by the overwhelming support for AI safety.
00:14You know, we've been doing a huge amount of surveying of not just the public, but investors, business leaders.
00:19And you see agreement across the board.
00:2187% of Americans are concerned about AI safety and want the president to take action on that.
00:27But they don't want to slow down. You know, we just released something called the National AI Monitor,
00:33which tracks how ordinary Americans are kind of experiencing AI in their lives.
00:38People are net positive on that, especially at work.
00:41So it's in this really interesting position where I think safety is the priority, but people like AI.
00:49They want to be trained on AI, positive experience, so no slowdown either.
00:54Do you see, though, this tension in that idea, that they want AI to advance, they want to be trained
00:59in it,
01:00and things like the data center backlash? Like, how do you kind of square those two ideas together?
01:05I think the data centers are a kind of manifestation of what AI could do.
01:11Like, it's a physical, you know, manifestation of that.
01:14But if you ask about people's experience of AI right now, positive in the workplace, that's, in fact, that's where
01:20people are most positive.
01:22Positive as consumers, positive about the impact of AI in their communities, but negative on the environment.
01:29Another interesting finding was the more people use AI, the more they're trained in AI, the more comfortable they get,
01:36and the more optimistic they get about the future.
01:38So I think that's a real lesson for business leaders, for example.
01:41Right.
01:42I also thought one of the interesting things you wrote about, so you had up to 30% of voters
01:47said that they would switch their vote to a candidate who aligns with their AI views.
01:51It reminded me, I had a conversation roughly about a year ago with the historian Neil Ferguson, who basically said,
01:562028, the main issue was going to be AI.
01:59And I thought it sounded crazy then.
02:00It's feeling increasingly less crazy, especially what we're hearing from the midterms.
02:04Do you think that this is the direction of things, that AI becomes a single voter issue topic?
02:08I do.
02:09I think that 30% is going to go up over time.
02:12And I think it's really incumbent on business leaders, investors feel the same way, all market stakeholders to come together
02:21and say,
02:22OK, how do we figure this thing out?
02:23You know, most of the world's airplanes are made by two companies, right, Airbus and Boeing.
02:27When you get on a plane, you expect it to not fall out of the sky and to work well.
02:32And so it's the same with AI.
02:33People want it to work well for them individually.
02:36They may be worried about what could happen.
02:38But for the most part, you know, certainly the AI developers and the AI deployers, who, by the way, are
02:46held equally responsible by the public if things go wrong,
02:49just incumbent on them to create the best experience possible.
02:53What do you make of the political divide in this amongst parties?
02:56That was interesting to me, too, that a very, very solid majority of Republicans want the U.S. to lead
03:01in AI,
03:02and less so from the Democrats feel that same way.
03:04I think it was something like a 28-point gap in those feelings.
03:06Yeah, the biggest political split was on whether America should be an AI leader.
03:11That was a 20-something percent split.
03:15But for the most part, Republicans, Democrats aligned with a few points on prioritization of safety,
03:23certainly aligned on how, you know, two-thirds of Americans would like the president to get a deal done with
03:30China.
03:31But even if he doesn't, they don't want no action on guardrails.
03:37So I don't think that split, when it comes to the big picture, is as great as it is.
03:41But it might show up in the way government acts on AI.
03:46If you were to advise – I know you're not a politician, and let's just couch that –
03:50but if you were to advise a congressional candidate right now based on this study,
03:53because this does feel like a goldmine for how the public is feeling, what would you advise?
03:58I would advise them to figure out, okay, what exactly does safety mean to the American people,
04:02and how can we work together with business and with the markets to create that future?
04:07Because once people buy in, you know, the big risk here, honestly, Danny,
04:11is if people lose trust and lose faith that AI is working for them and their family,
04:18you know, if they see fewer and fewer people getting wealthier and wealthier from AI,
04:22but their lives are worse, like, that's a real problem with our system of capitalism, democracy.
04:29And if people don't like the system, you know, they're apt to want to change it.
04:32So I think that's the key issue that any elected official has got to get their arms around.
04:37It's interesting, this idea of trust, and it's something I've been talking about,
04:40especially with Meadows, Mu's AI assistant, that it got billions of downloads –
04:45millions, rather, billions would be a lot – 25 – I think 2.5 million downloads
04:48in the space of something like a couple days. And I wonder what you make of that,
04:52that there is a cohort out there that clearly trusts this AI agent enough
04:57to give it its credit card data and its login information.
05:00Absolutely. People – so it's interesting. There's a Gallup poll out recently
05:04that people are very worried about AI, sort of an AI, you know, future, a doomer future.
05:11We don't see that in terms of individual experience. The more people use it themselves,
05:15you know, I think people will – they prefer their AI. Like, I like my AI,
05:20but I'm worried about AI's future impact on society.
05:24That's a really interesting dynamic, and it shows up in the Muse data.
