In determining why Kamala Harris and the Democrats lost so badly in this year’s elections, commentators have blamed various factors but missed several key reasons. Observers have cited the economy, immigration, threats to democracy, anger at elites and President Biden’s delay in leaving the race. These factors are important, but only partial explanations.
As a psychiatric researcher, I have conducted three decades of studies on people’s attitudes, and I see several problems with these analyses, particularly failures to fully appreciate the increasing, unique and powerful uses of AI and social media and their neurological impacts, and the limitations of survey methods to assess these forces.
Social media and AI are particularly well-suited to inciting emotion — especially if they use distortions and lies — and have thus radically transformed American politics.
AI facilitates the sophisticated fabrication of images, videos and voices. Each algorithm also continually learns and hones its abilities to determine which messages to send to whom — gauging what particular words and images get clicked by whom and then altering posts accordingly. AI-enabled emotional recognition and communication have been rapidly improving.
These new technologies thus make politics even more of a blood sport, uniquely provoking raw emotions of fear, hatred and violence, especially for candidates willing to use AI’s full capabilities for distortion. Social media is also replacing the roles of legitimate news services, which were generally committed to fair and balanced reporting. Social media also serves as a Trojan horse, letting Russia and China further undermine democracy.
Harris’s poll support was strongest immediately after her debate with Trump, in which his lies were most clearly contested and transparent. Once he refused to debate her again, he was able to fabricate his own versions of truth even more, and his polling improved. Reality became ever-less important.
Fifty-one percent of Trump voters, for instance, felt democracy was threatened, based on false claims that illegal immigrants were invading our nation and voting. No evidence of such voting exists. Yet Russian-fabricated video provided purported proof, which was disseminated virally on X. X owner Elon Musk has over 200 million followers (in comparison, only around 160 million people voted in the election), and actively sends MAGA posts virally and eliminates Democratic posts.
Harris provided detailed policies and concepts about democracy, appealing to intellect, but these proved insufficient against fake news. Republicans’ whipped-up emotion, proved far stronger.
Recent neuroscience research indicates why — different types of thinking involve different parts of the brain.
Researchers have devised, for instance, thought experiments involving a trolley car speeding down a track, about to hit and kill five people. Study participants are told to imagine they are standing on a bridge overlooking the track and could potentially shift a lever that would switch the trolley car onto another track instead, on which only one man is standing. He would get killed, but the other four people would consequently survive.
Other participants are told that next to them on the bridge is an overweight man whom they could push off the bridge. He would land in front of the trolley and stop it, killing him but saving all five people. While 87.1 percent of participants thought pulling the lever was morally acceptable, only 46.8 percent thought pushing the man was ok.
These two sets of decisions (pulling the switch vs. pushing the obese man) involve very different sections of the brain. The calculation that five deaths would be worse than one death is a more utilitarian decision, entailing cost-benefit analyses that occur in the dorsolateral prefrontal cortex in the front of the brain, near the neurological centers for reasoning and abstract thought.
Feelings about killing the obese man involve emotional and intuitive factors, empathy and personal interactions, which get processed in the so-called “reptilian” part of the brain that arouses us emotionally, responds to more emotionally gripping messages, and is far more deeply rooted in the middle of our skulls. The frontal cortex, involved in utilitarian calculations, evolved millions of years afterwards and matures much later in each of us.
Yet pollsters failed to adequately gauge these raw emotional, rather than rational-based voter decisions; and need to alter approaches. Election surveys commonly rely on forced-choice answers, such as, “Which of the following eight issues concern you most?” But such options did not include, “I don’t want a woman president” or “I don’t want a Black president.”
Voters might not directly admit to possessing sexist or racist attitudes, but in other studies, questions such as, “How much do you think ‘other people’ or ‘people you know’ do not want a woman?” reveal such attitudes. Pollsters could also include questions to detect so-called “social desirability bias” — for instance, “How concerned are you with other people’s opinions of you?”
Surveys also erred, in taking voters’ responses at face value, without probing the larger contexts, and how and why particular issues were felt important.
Voters’ “priorities” were based in large part on algorithm-enabled disinformation and emotions. It is not what votes knew, but what they felt they knew based on what disinformation they received. These “priorities” are therefore not necessarily what voters wanted based on facts and full information, but on AI-driven distortions.
Granted, historically, many voters have also not fully grasped key issues, but these new technologies are exponentially exacerbating these problems. Democrats may have used AI, but do not appear to have leveraged its unique abilities to foment emotion.
These phenomena have important implications for the future. Trump will likely be unable to fulfill all of his promises. Many are vague and contradicting. Such failures may not matter to millions of his supporters.
But to respond effectively, Democrats need to learn far more fully how to arouse and stir the brain’s emotional centers, developing strategies that far more effectively combat and use these new weapons, devising images and algorithms that continually refine how and to whom to target. This need is crucial since Elon Musk will continue using X's full force to push Trump’s agenda.
Pollsters need to adopt ways of better assessing these far more emotionally-driven trends. Otherwise, AI will win again.
Robert Klitzman, M.D., is a professor of psychiatry at the College of Physicians and Surgeons and the Joseph Mailman School of Public Health, and the Program Director of the Master of Science in Bioethics at Columbia University.