For more than two decades, groups concerned about the risks of artificial intelligence and its impact on society have warned that the creation of this technology may spell the end of humanity. Despite the increasing urgency of their rhetoric, their responses have been almost entirely technocratic, focused on technical research and. policy recommendations. The prophets of AI risk have built think tanks, not mass movements, much less militant cells.
That might be changing. This weekend, two apparent assassination attempts targeted OpenAI CEO Sam Altman’s home in San Francisco—the first high-profile anti-technology terrorism America has seen since the apprehension of the Unabomber. Whether more such incidents will follow is unclear, but they do reflect a widespread feeling that the forward march of AI has become unstoppable, even as many—including those within the industry—warn of its dangers. It is unsurprising that individuals have responded to such rhetoric with violent attempts to prevent the warned-of catastrophe. For these people, the expert-led approach to AI has run its course.
What would the spread of such violent resistance mean for the future politics of AI? In the 1960s, the social theorists Frantz Fanon and Hannah Arendt both published works grappling with the place of violence in social movements at a time when many revolutionary movements dedicated to decolonialization, racial justice, and class struggle were adopting violent methods. Their opposed views on these developments have a new relevance for the evolving AI debate.
In The Wretched of the Earth, Fanon argued that because colonial power had relied on violence to achieve dominance, the only way for the colonized to reclaim their agency was to take up the tools of violence available to them. Violence, for Fanon, was liberatory less because of its immediate effects than because it transformed the consciousness of the colonized. More than a tactic, violence was for Fanon a means of psychic empowerment, of moving from passivity to agency.
It isn’t hard to see how this logic might appeal to those who feel powerless in the face of runaway technological development. The seductive promise of direct action is the promise of exerting agency over the future, something many today feel that they lack.
“The seductive promise of direct action is the promise of exerting agency over the future.”
In her 1969 work On Violence, Arendt contested Fanon’s view, claiming instead that power and violence are opposites. Power, for her, was the human ability to act together, to form collectives that are robust and can pursue long-term goals; violence, in contrast, was the recourse of those unable to act within a wider collective, employed for short-term ends of disruption and rupture—a force that becomes unnecessary for those with true power. According to Arendt, goals such as national liberation and self-determination can only be achieved through non-weaponized power. That, she argued, was why the violent revolutionary movements and slave revolts lauded by Fanon have never resulted in the utopias they promised, but rather “turned dreams into nightmares for everybody.” Brutal coercion cannot build anything lasting.
Arendt would agree that the use of violence was a predictable development at a time when some Silicon Valley elites talk openly of humans being “obsoleted” by the technologies they are building. “Every decrease in power is an open invitation to violence,” Arendt wrote, “if only because those who hold power and feel it slipping from their hands… have always found it difficult to resist the temptation to substitute violence for it.” If violence is the recourse of the powerless, technologies of disempowerment will breed more violence.
She would, however, dispute the Fanonian idea that violence will empower its perpetrators. Violence, she believed, tends to beget more violence, as those who feel vulnerable as a result of its exercise against them become more likely to respond in kind. “The practice of violence, like all action, changes the world, but the most probable change is to a more violent world”—and therefore one in which fewer individuals have access to genuine power.
On the other hand, one circumstance Arendt foresaw in which violence could sustain power in perpetuity is in the case where power ceased to depend upon collectives of people. “Only the development of robot soldiers,” she said, which “permit one man with a push of a button to destroy whomever he pleased, could change this fundamental ascendancy of power over violence.” Let us hope that the non-violent exercise of collective power can ensure this button is removed from the hands of any one man.