How do Americans feel about AI?

We’re quickly coming up on one year since ChatGPT was released to the public. In that time, it and other generative AI tools have placed artificial intelligence front and center in a larger discussion about the future of work, art, ethics and pretty much everything else.

So, what do Americans think about AI now? The upshot is that many of you are checking the “somewhat concerned” and “mostly concerned” boxes on this one. And it seems like Democrats and Republicans are generally on the same page regarding the future of AI.

Marketplace’s Lily Jamali discussed the question with Ryan Heath, global tech correspondent for Axios, who recently combed through several surveys to get a sense of the country’s current sentiment toward AI.

The following is an edited transcript of their conversation.

Ryan Heath: Across a number of surveys that have happened over the last three or four months, Democrats and Republicans showed very consistent and skeptical attitudes towards a lot of AI. So that’s really unusual that we have a situation where Democrats and Republicans basically agree, even if perhaps they have different motivations for their fears and their concerns, and a situation where many people are skeptical of tech companies and how they’re handling AI, which leads to an unusual level of openness, let’s say, around having the government involved in this topic.

Lily Jamali: That’s really interesting. So this is not all that of a partisan issue. Talk us through where some of the dividing lines do lie. Are these differences of opinion on AI, are they falling mostly along generational lines?

Heath: That is absolutely top of the list, but it’s not only generational lines. So for example, we know that the younger you are, you’re much more likely to have used one of these new generative AI chatbots. And you’re probably more aware of AI that’s existing in the background in your life, whether that’s using a prompt service like Siri or having some kind of device in your home, like an Alexa device. For example, younger people tend to use it more often. At the same time, there are income divides. So the better off you are, the more income you’re earning, the more likely you are to have tried to use AI in the workplace, the more likely you are to have bought the devices that have AI in the background. And the more income you earn, that tends to be in the older age bracket. So we don’t see so many black and white divisions here. There are some racial divides as well. You have a very high level of interest and usage and experimentation among Asian American communities, but a lot lower level of awareness of some of these technologies among Black Americans. And the white population sits in the middle there. And interestingly, we don’t see so much of a gender divide. You know, little bits creep in here or there, but it’s single-digit differences in terms of how men and women are relating to the technology.

Jamali: What about education? How much does that determine whether people are on board here?

Heath: I think it’s not so much of a factor in terms of whether you have fears or concerns. But it definitely is a factor in whether you’ve actually used the technology. And in particular, more and more people are using it as part of their job in some way. About 40% of people who have a college degree have used this generative AI set of tools in the workplace. But only half that number who don’t have a college degree have used it in the workplace. And some of that is probably about the nature of the job. If you are working on a factory line or you’re serving food in a restaurant, maybe AI is there in the background, but you don’t really need a chatbot to do that job necessarily. But if you’re sitting at a desk all day, and you need to come up with a form of words or deliver a presentation, a chatbot is exactly the sort of thing that might save you a bit of time or help spark some ideas. And so that explains a bit of the difference in usage.

Jamali: So you describe this as a largely divisive issue right now, but divisive around what specifically?

Heath: Well, I think the biggest concern that people have is that this is going to upend their lives or their job in some way. So they understand that a set of machine learning can scan much more data than any individual human ever could in their lifetime. So if you’re somebody that needs to make sort of precision actions or take decisions based on a large amount of data, your job is at risk. And I think people intuitively understand that. And it doesn’t mean that everyone is going to be out of a job tomorrow. But I think people understand their jobs will be changing, some of their jobs won’t be there in the future. And people are also rightly concerned about how the big players in this field are going to use their information. You know, if your health data is circulating around on the internet, if your face has been scanned a million times, you want to make sure that the people who have that information are handling it correctly. And the big tech companies haven’t had a great track record on that front in recent years. And they’re also saying, we want the government to be involved. And the asterisk I would put on all of this is it’s easy to support a regulation in theory and to say you want somebody to get in and enforce the law and make sure it’s been done right. And then when people start to see the details of draft regulations, they might not be as impressed. You know, we might see those traditional fights happening in Congress where parties suddenly do discover their differences again.

Jamali: Right. For now, though, people are still figuring out what they think of all of this. How do you think sentiment around AI has changed from, say, a year ago, before tools like ChatGPT really went mainstream? Have you seen a shift?

Heath: Not a massive shift, I [would] say. But I think the other thing to know is that people weren’t doing polls about AI every two minutes, like they seem to be doing at the moment. So we’re getting much richer data than we used to have about this. And in the past, AI was even more remote. So, you know, we’re talking about how most Americans think they haven’t used AI yet or they haven’t used these new tools. When people were thinking about AI, they’re potentially thinking about some kind of fictional version of AI, or they do understand that, hang on, every time you unlock your phone by staring at it, that is a form of AI. So I think people’s links to AI are changing very rapidly. And so their opinions will evolve over time. But, you know, at the same time, there has been skepticism all along about these extremely powerful forms of technology.

Jamali: What do the results of your analysis mean for the companies that are developing this AI technology? Do they have to navigate this differently?

Heath: I just think the more sensible conversations you have about this and you have to invest some time and money in and yes, maybe the profits will be a little bit lower because you took another month and you sent another email and you checked with people another time about whether they were happy with it. I think when companies make that investment, they might start to see the trust go up because people are coming from this position of skepticism, and unless somebody invests more time in building up that trust, I think they’re going to stay skeptical for a while.

More on this

As Ryan Heath mentioned, many people seem to want some kind of government intervention or regulation for AI. That process is moving along, albeit at a snail’s pace.

The Federal Election Commission is considering asking the public for comment on regulating AI-powered deepfake material in political ads.

The issue was underscored when the presidential campaign of Florida Gov. Ron DeSantis drew criticism for featuring an AI-generated image of former President Donald Trump hugging Dr. Anthony Fauci, the former U.S. public health official whom many conservatives have vilified.

If the FEC moves forward, Americans would have two months to weigh in.