AI is choking democracy while also providing the tools for repression and manipulation to both authoritarian and democratically elected governments. But it's not too late to change the future of digital technology.
We cannot understand the political mess that social media has created without recognizing the profit motive based on targeted ads, which makes these companies prioritize maximizing user engagement and sometimes rage. Targeted advertisements, in turn, would not have been possible without the collection and processing of massive amounts of data.
The profit motive is not the only factor that has pushed the tech industry in this antidemocratic direction. These companies’ founding vision, which we dubbed the AI illusion, has played an equally important role.
Democracy, above all else, is about a multitude of voices, critically including those of ordinary people, being heard and becoming significant in public-policy directions.
The notion of the “public sphere,” proposed by the German philosopher Jürgen Habermas, captures some of the essential features of healthy democratic discourse. Habermas argued that the public sphere, defined as forums where individuals form new associations and discuss social issues and policy, is pivotal for democratic politics.
Using British coffeehouses or French salons of the nineteenth century as the model, Habermas suggested that the critical ingredient of the public sphere is the ability that it offers to people to freely participate in debates on issues of general interest without a strict hierarchy based on preexisting status. In this way, the public sphere generates both a forum for diverse opinions to be heard and a springboard for these opinions to influence policy. It can be particularly effective when it allows people to interact with others on a range of cross-cutting issues.
Early on, there was even a hope that online communications could generate a new public sphere, one where people from even more diverse backgrounds than in local politics could freely interact and exchange opinions.
Unfortunately, online democracy is not in line with the business models of leading tech companies and the AI illusion. In fact, it is diametrically opposed to a technocratic approach which maintains that many important decisions are too complex for regular people. The vibe in the corridors of most tech companies is that men (and sometimes, but not that often, women) of genius are at work, striving for the common good. It is only natural that they should be the ones making the important decisions. When approached this way, the political discourse of the masses becomes something to be manipulated and harvested, not something to be encouraged and protected.
The AI illusion thus favors an antidemocratic impulse, even as many of its executives view themselves to be on the center-left and supporters of democratic institutions and even the Democratic Party. Their support is often rooted in cultural issues and conveniently bypasses the vital building block of democracy: people’s active participation in politics.
Such participation is especially discouraged when it comes to AI because most entrepreneurs and venture capitalists believe that people do not understand the technology and unnecessarily worry about its intrusive effects.
As one venture capitalist put it, “Most of the fears of artificial intelligence are overblown if not altogether unfounded.” The solution is to ignore these concerns, forge ahead, and integrate AI into every aspect of our lives because “perhaps it’s only when a technology is fully integrated into daily life, and recedes into the background of our imagination, that people stop fearing it.”
This was essentially the same approach advocated by Mark Zuckerberg when he told Time magazine, “Whenever any technology or innovation comes along and it changes the nature of something, there are always people who lament the change and wish to go back to the previous time. But, I mean, I think that it’s so clearly positive for people in terms of their ability to stay connected to folks.”
Another aspect of the AI illusion, the elevation of disruption as a virtue encapsulated by “move fast and break things,” has accelerated this antidemocratic turn.
Disruption came to mean any negative effects on others, including workers, civil society organizations, traditional media, or even democracy. It was all fair game, in fact encouraged, so long as it was a consequence of exciting new technologies and consistent with bigger market share and moneymaking.
A reflection of this antidemocratic impulse can be seen in Facebook’s own research on how users respond to negative and positive emotions from friends in their newsfeed.
In 2014, the company undertook a massive internal study, manipulating the newsfeed of nearly seven hundred thousand users by reducing their exposure to either positive or negative expressions for a week. Unsurprisingly, greater exposure to negative emotions and lower exposure to positive emotions impacted users, with lasting adverse effects.
The company did not ask for any permission for this massive study from the users or even attempt to adhere to commonly accepted standards in scientific research, where informed consent from subjects is necessary.
After some of the results of the study were published by Facebook researchers and others in the Proceedings of the National Academy of Sciences, the editor in chief published an Editorial Expression of Concern because the study was done without informed consent and did not meet accepted standards of academic research.
Google followed the same playbook in its efforts to expand the amount of information that it collected with Google Books and Google Maps. The company ignored privacy concerns and acted first and without permission or consultation, hoping that things would get sorted out or, at the very least, its fait accompli would be accepted. That worked out, at least for Google.
Facebook and Google are not exceptional in the industry. It is now routine for tech companies to collect vast amounts of data without any consent from the people whose information or photos are being harnessed.
In the area of image recognition, for example, many AI algorithms are trained and sometimes take part in competitions on the ImageNet data set, initiated by the computer scientist and later chief scientist of Google Cloud, Fei-Fei Li.
The data set, which contains more than 15 million images sorted into more than 22,000 categories, was built by collecting private photos uploaded to various applications on the internet, with no permission from the people who took or appear in these pictures. This was generally viewed as acceptable in the tech industry. In Li’s assessment, “In the age of the Internet, we are suddenly faced with an explosion in terms of imagery data.”
According to reporting in the New York Times, Clearview has systematically collected facial images without consent, aiming to build predictive tools that identify illegal immigrants and people likely to commit crimes. Such strategies are justified by arguing that large-scale data collection is necessary for technological advancement.
As an investor in a facial-recognition start-up summed up, the defense for massive data collection is that “laws have to determine what’s legal, but you can’t ban technology. Sure, that might lead to a dystopian future or something, but you can’t ban it.”
The truth is more nuanced. Imposing massive surveillance and data collection is not the only path of technological advance, and limiting it does not mean banning technology. What we are experiencing instead is an antidemocratic trajectory charted by the profit motive and the AI illusion, which involves authoritarian governments and tech companies foisting their vision on everybody else.
Democracy Undermined When We Most Need It
The tragedy is that AI is further undermining democracy when we need it most. Unless the direction of digital technologies is altered fundamentally, they will continue to fuel inequality and marginalize large segments of the labor force, both in the West and increasingly around the world. AI technologies are also being used to more intensively monitor workers and, through this channel, create even more downward pressure on wages.
You can pin your hopes on the productivity bandwagon if you like. But there is no indication that shared productivity gains will be forthcoming soon. As we have seen, managers and entrepreneurs often have a bias to use new technologies to automate work and disempower people, unless reined in by countervailing powers. Massive data collection has exacerbated this bias.
Countervailing powers are hard to come by without democracy, however.
When an elite completely controls politics and can use tools of repression and propaganda effectively, it is hard to build any meaningful, well-organized opposition. So robust dissent will not rise in China anytime soon, especially under the increasingly effective system of censorship and AI-based surveillance that the Communist Party has established. But it is also becoming increasingly difficult to hope for the resurgence of countervailing powers in the United States and much of the rest of the Western world.
AI is choking democracy while also providing the tools for repression and manipulation to both authoritarian and democratically elected governments.
As George Orwell asked in 1984, “For, after all, how do we know that two and two make four? Or that the force of gravity works? Or that the past is unchangeable? If both the past and the external world exist only in the mind, and if the mind itself is controllable, what then?”
This question is even more relevant today because, as philosopher Hannah Arendt anticipated, when bombarded with falsehoods and propaganda, people both in democratic and nondemocratic countries stop believing any news.
It may be even worse than that.
Glued to their social media and frequently outraged and very often absorbed by strong emotions, people may become divorced from their community and democratic discourse because an alternative, segregated reality has been created online, where extremist voices are loudest, artificial echo chambers abound, all information is suspect or partisan, and compromise has been forgotten or even condemned.
Some are optimistic that new technologies, such as Web 3.0 or the metaverse, can provide different dynamics. But as long as the current business model of tech companies and the surveillance obsession of governments prevail, they are more likely to further exacerbate these trends, creating even more powerful filter bubbles and a wider wedge with reality.
It is late, but perhaps not too late.
The Future Path of Technology Remains to Be Written
The tech industry and the large corporations are politically more influential today than they have been for much of the last hundred years. Despite scandals, tech titans are respected and socially influential, and they are rarely questioned about the future of technology—and the type of “progress”—they are imposing on the rest of society. A social movement to redirect technological change away from automation and surveillance is certainly not just around the corner.
All the same, we still think the path of technology remains unwritten.
The future looked bleak for HIV/AIDS patients in the late 1980s. In many quarters they were viewed as perpetrators of their own fate, not as innocent victims of a deadly disease, and they did not have any strong organizations or even any national politicians defending their cause. Although AIDS was already killing thousands of people around the world, there was very little research for a treatment or a vaccine against the virus.
This all changed during the subsequent decade.
First there was a new narrative, showcasing the plight of tens of thousands of innocent people who were suffering from this debilitating, deadly infection. This was led by the activism of a few people, such as playwright, author, and film producer Larry Kramer and author Edmund White. Their campaigns were soon joined by journalists and other media personalities.
The 1993 movie Philadelphia was one of the first big-screen depictions of the problems of HIV-positive gay Americans, and it had a major impact on the perceptions of moviegoing audiences. TV series tackling similar issues followed.
As the narrative changed, gay-rights and HIV activists started organizing. One of their demands was more research into cures and vaccines for HIV. This was initially resisted by US politicians and some leading scientists. But organizing paid off, and soon there was an about-face by lawmakers and the medical policy establishment. Millions of dollars started pouring into HIV research.
Once the money and societal pressure built up, the direction of medical research altered, and by the late 1990s, there were new drugs that could slow down AIDS infections, as well as novel therapeutics, including early stem-cell treatments, immunotherapies, and genome-editing strategies. By the early 2010s, an effective cocktail of drugs was available to contain the spread of the virus and provide more normal life conditions for most infected people. Several HIV vaccines are now in clinical trials.
What seemed impossible was achieved fairly rapidly in the fight against HIV/AIDS, as it was in renewable energy. Once the narrative changed and people became organized, societal pressure and financial incentives redirected the path of technological change.
The same can be done for the future direction of digital technologies.
Adapted from Power and Progress: Our 1,000 Year Struggle Over Technology & Prosperity. Copyright © 2023 by Daron Acemoglu & Simon Johnson.