Understanding Anti-AI Public Opinion

People can accept tradeoffs when they see value — but if they don’t, what happens?

Photo by Elena Mozhvilo on Unsplash

Public Opinion

There has been a lot of public discourse in the last few weeks focusing on the issue of public perception of AI, and why it’s so negative . People polled indicate they hate AI more than ICE , and younger generations seem to have stronger negative opinions , even as many of them use the technology day to day.

Many journalists and politicians are looking for a single factor to explain this pattern. Is it the power bills? AI slop taking over our social media feeds? The irritating noise data centers make? Chatbots confidently asserting stupid things? AI executive billionaires jetting around the world while grocery prices rise? City governments giving away local resources to corporate interests without considering or consulting their constituents? Attempting to simplify people’s complex opinions in this way is counterproductive.

The question is timely because the tech industry is attempting to build data centers across the country to increase AI compute capacity, and people are protesting , loudly . My argument is that negative opinions are largely driven by negative personal experiences with AI. That probably wouldn’t be sufficient by itself to inspire a protest movement, but combine that with practical, material costs AI is inflicting on communities and society, and the lack of practical value it’s generating, and it’s easy to see why these protests have developed. Why should we allow major infrastructure into our community to power a technology we think is harmful?

The truth is, and I think we all intuitively know this, there’s not just one thing driving public opinion. If you pile up all the information you’re receiving, from your own lived experience, friends, family, news sources, social media, etc, these are the raw materials that combine, along with your upbringing, religious training, education, etc, to make your perspective. We all work this way when developing our opinions and beliefs. What you personally witness and experience has a major impact on where you end up.

This is part of why pundits fret about “ why American consumer sentiment is the worst it’s ever been, even though the macro metrics of the economy indicate no recession” — our perceptions are more than just the composite of a set of macro-level data. Events happen to us, or to our friends and loved ones, and we interpret those phenomena when we create our broader understanding. We see the gas prices and grocery prices with our own eyes. We hear we aren’t getting raises this year, or that layoffs are looming for us or our loved ones. We associate things together because we are creating a cohesive universe to help us understand and cope with what is going on. In the current moment, Americans are struggling to understand and cope with a lot.

Experiencing AI

People are gradually accumulating personal experiences and information about AI, and they’re finding that the overall impact feels more harmful than beneficial. People have lots of opportunities to interact with AI, there’s no shortage of practical experience with it, but the results are just not positive a lot of the time. They may directly use it, in the form of chatbots for example, or they may witness its effects, such as seeing AI slop in social media feeds. We could discuss endless examples of this, from AI in education and childhood development to AI in health, from AI affecting the environment to AI and government.

For one example, let’s take AI in the workplace. In the workplace, a very dominant narrative is “AI will take your job”. White collar workers in particular are experiencing layoffs that bosses attribute to AI . (Many of these layoffs are actually attempts by corporations to bump stock prices rather than the result of AI successes. However, the corporations blame AI so that’s the narrative that gets created.) People are also being told by management, “If you don’t use AI enough, you’ll get fired”, but nearly simultaneously “If you use AI too much, the business will go bankrupt”. People feel how hypocritical and bizarre this is .

There are positive stories, but they have downsides. “AI will make you more efficient and enable you to do more,” rings hollow to people fearing for their jobs and fearing that the demands made of those who remain will continue to rise. Many types of work are unaffected at all, as well, such as many kinds of blue collar and care work. We hear “AI will take over art, design, ads, television, movies, and music,” and while this might sound good to people who don’t have the creative skill themselves, creatives see disaster unfolding for their economic prospects. Workers are set up with AI as their enemy, and any efficiency from AI is perceived as zero-sum with workers’ jobs. Business being more efficient is perceived as benefiting corporations and billionaires, not individual workers.

The negative AI experiences and negative feelings about billionaires have a strong connection. If you encounter AI and have a bad experience, you’re likely to wonder “why did this happen to me?” and may begin to inquire about the AI industry. Asking who benefits from AI being inserted into your workplace, or your doctor’s office, or a data center being built in your neighborhood, can lead to the doorstep of billionaires who are pushing AI investment and adoption across the economy, and who seem to have very little awareness or understanding of (or interest in) the lived experience of average people. This breeds resentment, and now we’ve drawn a pretty direct line between your negative experience and the actions of billionaires . When billionaires then come proposing data centers and AI investment in your community, the sentiment is strong.

It’s worth reminding ourselves that AI isn’t all bad. In fact, AI has many positive applications — software engineers are seeing it make a significant impact on our productivity and efficiency at work. Radiologists are using AI to improve the quality of scan readings. There’s a long list of helpful things AI can do. The problem is that the benefits frequently have a downside, especially for individuals, and often the downsides overwhelm people’s perceptions of the benefits that accrue in the aggregate.

Data Centers

A large part of why the media is so interested in the strong negative reactions to AI right now is because it’s manifesting in local action against data centers. Not all data centers even have any relationship to AI, but the two have been associated in people’s thinking, and anti-AI sentiment is driving people to protest locally and demand moratoria on building data centers.

Why data centers, and why now? Because people have developed these negative personal experiences with AI and with data centers built inconsiderately, and this is the action available to them. Government has broadly abdicated responsibility for regulating AI, although a few individual states are attempting to fight that battle. Most people perceive their elected officials as having little leverage to protect constituents from AI’s negative effects or externalities. However, when it comes to data centers, local permitting and regulation are required. Citizen complaints and protests can have an impact on the decisions of municipal and county officials in a way that is much more difficult for state and federal cases.

There are several arguments against data centers that have nothing really to do with AI as a technology. Consider the use of electricity, which might not matter to many people, except that the power infrastructure in America means that creating new power capacity has costs that are passed on to consumers. While we’re not being made to pay for the specific kilowatt hours that the data center uses, we’re being made to pay for building the power plants and lines that produce and transmit it. There are also real concerns about water usage, as well as displacement of wildlife and noise pollution. However, we have a lot of infrastructure in our communities that uses resources and has environmental downsides. We do not have large, organized protest movements preventing towns from building new Wal-Mart stores. The general public is not demanding that we stop growing almonds, which consume tremendous amounts of water, or eliminate cattle farming, which has significant local environmental impacts. If people had experiences with Wal-Mart and almonds as negative as those they have with AI, and no upside, that might be different. But for AI, people really aren’t seeing the positive tradeoff that would be required to accept the costs.

What Happens Next?

So, where does this leave us? AI is simply not demonstrating value to people commensurate with the costs they see. In fact, AI companies are accumulating serious negative baggage in the minds of the public, by continuing to fail to manage the harms AI can cause, and not demonstrating compelling benefits to regular people.

It’s a mistake to decide that people must just not understand AI, and to blame their opinions on ignorance. We know many people have tried AI , based on usage statistics and polling, and even among those who haven’t, it would be difficult to have no exposure to AI slop online in this day and age.

It’s also a mistake, I think, to determine that the environmental and social impact of data centers is a standalone reason that opposition has gained influence. We have many, many things that we do at all levels of society that are environmentally dangerous or socially harmful. We build superhighways through neighborhoods, we build coal power plants, we built airports with flight paths over homes. All of these have costs that we have debated and determined are acceptable because of the benefits we gain, both individually and communally, from these structures. What makes data centers different? People don’t think the theoretical or actual benefits are sufficient to outweigh the costs, and until the AI industry changes that, public opinion isn’t going to change.

Read more of my work at www.stephaniekirmer.com

Further Reading