If you’re most interested in the social effects of artificial intelligence, you’re in a more subjective playground than the quants who are more tuned in toward advances in hardware or token cost. You’re looking at trends that are, in some ways, pretty amorphous, and shifting tides that can be difficult to identify.

We know intuitively that many people distrust AI. But why? Is it because of job displacement, or something else? And how big will the cycle of job displacement be? How many jobs will be ultimately lost to AI?

These are largely unanswered questions at this point. We don’t know exactly how the biggest tech transformation of our lifetime is going to play out. Recent BLS stats show jobs are dropping – but is that because of AI?

Pushing Back

One major concern is that people are going to “fight back” against AI basically out of their own self-interests.

“AI might get a lot more human-like, or we might just not care that it’s only simulating something like empathy,” says NPR Correspondent Sally Helm, in a piece that focuses on the impact of AI on jobs, and reactions from a beleaguered workforce.

But there’s also another kind of anxiety around the capabilities of LLMs: namely, that with the power of AI to create misinformation, a right-wing populist movement may be able to use it for campaigns that are the stock in trade of authoritarian models.

Here’s how Clifford Young puts it at the Center for International Governance Innovation:

“In the next decade and beyond, our caudillo syndrome, together with AI’s disruptive nature, will only make governance that much more difficult. Here, my research leads me to believe we are entering a ‘populist super-cycle’ that could last years, as belief that the system is broken continues to trend upward. We should expect increasingly more contested elections around the world, further sowing the seeds of distrust.” (For reference, with “caudillo syndrome,” Young is talking about a concept of a strong leader retrieving power from an order that doesn’t seem to have popular backing.)

There was a lot more on this from a panel at our Imagination in Action conference at Stanford, titled AI and the Future of Work. This came up.

Worries from Both Sides

About a decade ago, I got really interested in the rise of populism,” said Noreena Hertz, an author on AI and jobs. “Why were people voting, across the world, for these right-wing populist leaders? I started researching it, and one of the key drivers was, it turned out, automation.”

A lot of this fear, she indicated, is still pretty theoretical.

“What was really interesting was, it wasn’t necessarily that people voting for right-wing populists had had their jobs taken away by automation, but they were very scared that their jobs would be taken away by automation. So when I think about what’s happening now, and where we’re at today, one of the things that’s on my mind is, are we going to see a populist uprising? With AI as its focus? … a populist uprising on the left, not the right. Because what makes this potential set of job dislocations very different, of course, is that it’s white collar workers who are likely to be dislocated, rather than blue collar workers.”

Hertz spoke to what she called a “cadre” of left-wing political leaders speaking to people about AI.

“They are already speaking about this very clearly, when you look at their rhetoric about AI and jobs. So are we moving into a situation where, if we do nothing about it, where, instead of, let’s say, immigrants being the ‘other’ and being under attack by populists, it will be people here, the tech elite, facing the ire?”

That was by no means the sole focus of the panel. Some pretty great people in AI talked about opportunity, too, and about how to gird the workforce with what it needs to thrive.

But we have to look at the challenges, too. We have to look with clear eyes at human response, at how human workers will be affected. For instance, there’s data out there suggesting that women will be less served by tools that allow workers to skill up and manage AI, which panelists agreed workers will need to do.

More Quotes from Stanford:

“The challenge is actually the context piece. If you look at all of these organizations, if you take the smartest PhD student from Stanford and you put them into the sales team at one of these companies, they’re not going to be successful, not because they’re not smart, but because they don’t have the context. And so a lot of what I think about right now is, how do we actually take that contextual knowledge an organization has, and bring it into their agent building process?” – Nancy Xu

“I’ve recently done some interviews with students to see what they’re worried about, and they’re really worried about jobs and not getting a job. But one thing that kept on coming up was they’re also really worried that they’re not going to get a fair shot at an interview – the way that the whole interviewing process is changing, … they’re worried that everyone’s using AI to create their resumes and create their backgrounds, and then the resumes are being read by AI, and they can’t work out how to game the system, and that … the people who can afford the pro version and pay for the subscription will then have an advantage.” – Noreena Hertz

“I actually think non-technical individuals have a lot of power in the world of AI. I think it’s not just about the individuals who know how to build the foundation models, but what I’ve observed is, I think actually a lot of the knowledge, and going back to that context question, a lot of important knowledge for making agents successful lies with non-technical developers.” – Nancy Xu

All of this seems really important for the future, as we see 2025 wind down. Stay tuned for more.