The sudden ubiquity of ChatGPT, a form of artificial intelligence (AI) that one would have relegated to science fiction merely a few years ago, means this powerful tool is now accessible on every computer with an internet connection. Given its capacity to create, destroy, and transform, we wanted to know what 14 Pennsylvania swing voters thought of the technology.
We asked seven Republicans, four Democrats, and three Independents—all Trump-to-Biden voters—about their views on AI when we focus grouped them on April 11. While all said they were familiar with the term, there were widely divergent attempts to define it. Ten had heard of ChatGPT, and four said they had used it. One created a presentation, which he then transferred to PowerPoint, while two others experimented with it to see what it could do.
A fourth person used it to write a story. “It scared me. It was impressive. I could see where people would get a little freaky about it,” said Donna, 57, from Altoona, PA.
To give people a sense of what the technology can do, we performed a few demonstrations of ChatGPT. First, we asked ChatGPT to compare, in 150 words or less, two professional sports teams in Pennsylvania in notable seasons for those franchises (e.g., the 2008 Pittsburgh Steelers vs. 1980 Philadelphia Phillies). In a matter of seconds, the program composed a human-sounding compare-and-contrast mini-essay describing the key players and other details about each team.
It just so happened that the focus groups took place the day after the 53rd anniversary of the Beatles breaking up in 1970. This inspired us to ask ChatGPT a highly speculative question: “Based upon the actual songs released by former Beatles in 1971, if the Beatles had remained together in 1971 and put out one album, which 12 songs would be on that album?”
The program immediately produced a playlist of 12 songs, including “Imagine” by John Lennon, “Maybe I’m Amazed” by Paul McCartney, “All Things Must Pass” by George Harrison, and “It Don’t Come Easy” by Ringo Starr.
In a third demonstration, we instructed ChatGPT to turn that list of 12 songs into a poem—again, just in a few seconds. It read, in part:
“In ’71, the Fab Four were done,
But if they’d stayed, and worked as one,
What twelve tracks could they have spun?
Imagine, by John, so bright and bold,
A world at peace, and love untold”
In one session, where the respondents skewed middle-aged, their reactions included “amazing,” “it’s impressive,” “I’m shocked,” and it’s “something I didn’t really think I would see in my lifetime.”
In the other session, where the respondents were a bit younger on average, their comments were less enthusiastic.
“My impression of the AI stuff is really it’s just supposed to simplify work processes. It’s not going to be the next Shakespeare. It’s not going to make art that’s going to last through time,” commented Megan, 34, from Halifax, PA.
“I like the technology from what I see,” remarked Alfred, 41, from Philadelphia. “From what I read, there’s still a lot of flaws in it.”
The respondents were split on whether AI poses more promise or danger. Those who believe it offers more promise explained that we have seen a lot of technological advancements with computers, and we’re smart enough to be able to control it and benefit from it in the future. It has “unlimited potential” as a “useful tool” to help society. For example, one respondent remarked that it could help to detect a tumor sooner than a doctor would.
Those who say it poses more danger cited Elon Musk’s call for a six-month pause on the development of AI. They believe the technology is ahead of where regulations are at this point, suggesting it could be misused and pose risks to society. They want protections in place before we fully engage with the technology. Also, they’re concerned it could eliminate the need for the jobs they and people they know do.
To put a fine point on their concerns, half thought tech companies can get it right when it comes to limiting AI’s negative impacts, but significantly, none of the 14 thought these firms will get it right.
When we showed a slide outlining where possible job loss might occur due to AI advancements, such as in office administration, law, accounting, and graphic design, nine said they were concerned AI could replace the job of someone they know.
“I work in supply chain management,” commented Katina, 50, from Audubon, PA. “A good AI could do my entire job, so not just me, but my entire team. As a 50-year-old female, it’s a little scary. There’s not a lot of jobs out there once you start hitting that over-50 age range.”
Mike, 41, from Bridgeport, PA, remarked, “I am a paralegal…If [AI] can take huge chunks of data and churn out a memo and supporting documents in 30 seconds, why would you spend money on someone when that’s going to take them at least 45 minutes, an hour, or more? That is alarming.”
Based upon what we heard, there’s likely to be much more alarm—as well as joy—over AI advancements in coming years. In these early days, we’re still largely guessing how the public will react if/when major transformations occur, and anticipating which entities will feel the inevitable backlash.
Rich Thau is the president of the research firm Engagious, which specializes in message testing and message refinement for trade associations and advocacy groups. He is also the moderator of the Swing Voter Project, conducted in partnership with Sago. Matt Steffee is vice president of research services at Engagious.