For those of us growing weary of the predictions made by Silicon Valley CEOs about artificial intelligence possibly leading to the extermination of the human race, the new Israeli documentary NAZA is a counterpoint, even if that wasn’t the filmmakers’ intention.
The nearly 80-minute film takes us far away from the tiresome tech doomers (who keep racing ahead with the technology despite their apocalyptic concerns) and transports us to the nighttime rooftops of Tel Aviv, where Israeli filmmakers Yuval Abraham and Rachel Szor interview 24 anonymous members of the Israel Defense Forces (IDF) about their role in the relentless military strikes on Gaza over the past three years.
For those of us growing weary of the predictions made by Silicon Valley CEOs about artificial intelligence possibly leading to the extermination of the human race, the new Israeli documentary NAZA is a counterpoint, even if that wasn’t the filmmakers’ intention.
The nearly 80-minute film takes us far away from the tiresome tech doomers (who keep racing ahead with the technology despite their apocalyptic concerns) and transports us to the nighttime rooftops of Tel Aviv, where Israeli filmmakers Yuval Abraham and Rachel Szor interview 24 anonymous members of the Israel Defense Forces (IDF) about their role in the relentless military strikes on Gaza over the past three years.
NAZA, which takes its name from the Hebrew acronym for “collateral damage,” focuses heavily on the AI systems that became an integral part of the war in Gaza after the Hamas attack of Oct. 7, 2023. “‘You have to trust the system,’” one soldier recalls an officer telling him. Otherwise, “‘everything will fall apart.’”
The soldiers talk about the role they played in Israeli programs such as Lavender, an AI system that identifies enemy targets, and the unfortunately named Where’s Daddy, which involves using children’s cellphones to track when a Hamas target—their father—returns home. In the case of Where’s Daddy, the resulting strikes, as one soldier recounts, would kill the suspected Hamas member but typically also his family, including the child whose phone was hacked.
Some of the concerns expressed by soldiers in NAZA are not new, such as modern warfare being akin to a video game, a sentiment frequently expressed since the advent of armed drones and push-button warfare. In fact, there are no major revelations in the film—the filmmakers published many of the details about these AI systems previously in the Guardian, which co-produced the film; +972 Magazine; and the Hebrew-language outlet Local Call. But that’s perhaps beside the point: The immense power of the film is derived from the firsthand testimony of the soldiers who took part in this AI-driven system of destruction and now wrestle with the consequences of their actions, including many thousands of dead civilians.
So, what then does their testimony tell us about AI and war? In NAZA, no one is worrying about rogue AI killing us off—that’s more a concern for Silicon Valley companies looking to raise capital. Rather, the film is rightfully concerned about the way that AI allows the military to justify killing ever more people, even as these systems absolve humans of the moral responsibility for choosing targets.
“My role is entirely technical,” says one IDF member, describing his part in the Gaza war.
In this sense, AI gives soldiers a certain remove from the decision-making and even imbues such systems with virtue, since machines aren’t driven by hatred. “Having AI decide who lives and dies is better than leaving the decision to someone whose friend was killed by Hamas last week,” a soldier says.
Yet the film quickly reveals the hollowness of this sentiment, since we learn that in targeting a Hamas operative, the Israeli military sometimes knowingly allowed dozens of noncombatants to be killed as collateral damage—as many as 50 civilians per strike.
Yet if AI is supposed to transfer responsibility for killing from humans to computers, and thus absolve humans of moral responsibility, it doesn’t quite work. The humans who appear in NAZA clearly feel responsible—or at least responsible enough to risk their careers and potentially their freedom to give interviews to the filmmakers. (Israeli Prime Minister Benjamin Netanyahu has threatened to strip the filmmakers of their citizenship.)
Even as NAZA is, in some respects, an indictment of AI, in the end it’s more an indictment of humans and, in particular, humans born in a country traumatized by the Holocaust and then traumatized again by the attack of Oct. 7. On that day, fighters from Hamas and other groups brutally killed some 1,200 people and abducted another 251, holding many of them hostage for months or years. The film’s message is that Israelis, the victims of so much suffering, lost their moral core that day and created a system that has allowed for the killing of extended families and the destruction of entire city blocks, including hospitals and schools, in pursuit of Hamas militants.
The war has also clearly led to another form of trauma for at least some Israelis: soldiers who lament their role in a campaign that has killed at least 74,000 Palestinians in Gaza, according to numbers tracked by local health authorities. One Israeli soldier interviewed in NAZA describes hearing the screams of children through his headphones as they were killed in airstrikes. Another says he saw video of stray dogs consuming dead bodies in Gaza.
The anonymous Israeli soldiers who appear in the film understand a complexity that is difficult for many observers of this polarizing conflict—those partial to one side or the other—to absorb: that the horror of Hamas’s attack, which did not distinguish between soldiers and civilians, was followed by another horror—an Israeli war whose operators ostensibly targeted combatants but also knowingly civilians.
Where does this leave us with AI and warfare? Can AI actually improve the way wars are fought, or will we use it merely to absolve ourselves from blame when innocents are killed? And turning back to the doomers wringing their hands while simultaneously fundraising to build AI faster: If AI is so powerful, why then, after three years of war, is Israel still carrying out strikes against Hamas in Gaza?
That last question reminds me of an interview I did with a Pentagon scientist at the height of the war in Afghanistan, when the military was using advanced algorithms—the predecessor of AI—to predict insurgent attacks on U.S. and coalition forces. In a moment of frustration, the Pentagon official admitted that the Taliban appeared to be holding their own in the war, without fancy algorithms or even computers.
A decade after we spoke, the United States pulled out of Afghanistan, allowing the Taliban to quickly seize power. Now, the computers have grown exponentially better, AI is generating targets, yet U.S. and Israeli victories remain elusive, whether in Gaza or now in Iran.
As for the morality question, there is so far no evidence that AI is better at selecting military targets and avoiding civilian deaths, only that it chooses targets faster and that it gives the veneer of legitimacy. That perhaps is the inadvertent message of NAZA: It’s not that AI is all powerful but that it provides a technocratic justification for strikes that often end up killing innocent people.
Hollywood long ago recognized this folly, even when it was theoretical. In many movies about AI, the real danger is not necessarily the computers devoid of empathy that lead us to near self-destruction but the deluded humans who gave them that power in the first place.
In NAZA, the AI systems are the tools of death, but the real problem is the humans.
In the end, AI may or may not kill us all, as the CEOs of companies rushing to build the technology warn us. Or, as NAZA demonstrates, it may just spit out more names than humans could possibly gather on their own and let us do the killing ourselves. We seem to be good at it.