On Nov. 14, 2001, five weeks into America’s war against al-Qaida, a small, unmanned, remote-controlled airplane called a Predator took off from a U.S. air base in Uzbekistan, crossed the border into Afghanistan, and—with a video camera attached to its belly—started tracking a convoy of vehicles believed to be carrying jihadi leaders along a road in Kabul. A group of officers and spies, monitoring the streamed footage from inside a trailer in a parking lot at CIA headquarters in Langley, Virginia, watched the convoy stop outside a building. With the push of a button, the Predator fired a Hellfire missile at the building, the back half of which exploded. Seven people, survivors of the blast, were seen fleeing to another nearby structure. A second Hellfire destroyed that shelter, too. Among the dead was Mohammed Atef, al-Qaida’s military chief and Osama Bin Laden’s son-in-law. Five weeks earlier, on Oct. 7, a drone strike had been launched against another caravan, this one carrying the Taliban leader Mullah Omar, but the missile missed; doubts about this newfangled technology remained. But now, after the Atef killing, the era of the armed drone—the weapon that has since come to define American-style warfare in the 21^st century—had unambiguously begun.*

It’s an unlikely story how this weapon, and this era, came to be. The idea had been hatched all the way back in the early 1970s, the brainchild of the Pentagon’s chief scientist at the time, a nuclear physicist named John S. Foster. A model-airplane enthusiast, Foster envisioned loading a somewhat larger version of his hobbyist toy with a ground-scanning camera that could send back real-time images to a command post and a bomb that could be released by remote control. None of this technology existed yet, but as a crude experiment, Foster commissioned the Defense Advanced Research Projects Agency to build two “remotely piloted vehicles,” each powered by a lawn-mower engine and capable of staying aloft for two hours while carrying a 28-pound payload.

Foster’s experiment coincided with two developments in the real world. The first was the microprocessor revolution, which created the possibility of what came to be called “smart bombs”—weapons, either fired by missiles or dropped by planes, that could explode within a few feet of a target, thus allowing specific objects to be destroyed without doing much damage to the surroundings.

The second trend was the Soviet Union’s growing military strength. The USSR and its Warsaw Pact allies had long outnumbered American and NATO armies on the ground in Europe, but the United States compensated with nuclear supremacy. By the late 1960s, the Russians had achieved “nuclear parity”: If the United States launched a nuclear attack on Russia, the Russians could respond in kind. In order to deter and repel a Soviet invasion of Western Europe, NATO would need a qualitative edge in non-nuclear weapons. Foster’s souped-up model airplane seemed to fit the bill.

The idea was that, in the event of a invasion, these remotely piloted vehicles (much smaller and thus harder to detect than manned combat planes) could destroy targets well behind enemy lines—knocking out air bases, supply depots, follow-on echelons of tanks and other armored vehicles—and could thus disrupt and delay the Soviet offensive, giving NATO a chance to regroup and fight back, without having to fire back with nukes.

Prototypes were developed in the 1980s but few officers evinced any interest in the project until the 1991 Gulf War, when

U.S. Air Force and Navy planes dropped the first smart bombs, which were guided to their targets by laser beams. There weren’t as many smart bombs as cable-news coverage of that war suggested (they comprised only about 9 percent of all the American bombs dropped), and quite a few of them veered off course (a fact not known till well after the war). But some Pentagon officials began to talk about a “revolution in military affairs” (even calling it by its acronym, RMA), in which the qualitative edge provided by technology—super-accurate weapons, super-fast data transmission, and the ability to connect the two—would transform the nature and pace of warfare. The smart bombs were different from Foster’s vision—for one thing, they were dropped from manned aircraft—but they relied on similar technology, so drones too fell into the discussion of RMA.

In the spring of 1996, William Perry, President Bill Clinton’s secretary of defense, approved production of the first remotely piloted drone, the Predator RQ-1, and placed it in the hands of the Air Force. The “R” in RQ-1 stood for reconnaissance: For the moment, this drone was purely an intelligence tool, transmitting video footage back to base, where a “pilot” would steer its course with a joystick and analysts would scrutinize its images. In 1999, these new weapons provided vital intel about targets and troop movements for NATO’s air war in Kosovo, but the drones didn’t carry, much less fire, any weapons.

Around the same time, al-Qaida started looming as a serious threat. CIA Director George Tenet and White House counterterrorism chief Richard Clarke argued that Predators should help track down Osama Bin Laden in Afghanistan. The first flights for this mission occurred in October 2000. The Predators’ video feeds were beamed back to a monitor at CIA headquarters. Clarke recalls his first viewing as something out of science fiction, “this idea that I could tell someone ‘Can you move it a little to the left’ and, halfway around the world, something moved to the left.”

On one of the video feeds, a tall, bearded man could be seen. The image wasn’t sharp; in fact, it was fuzzy, but CIA counterterrorism analysts guessed—though they couldn’t be certain—that this tall man was Bin Laden. This inspired Clarke, then others, to push for putting a missile on the Predators so that the next time they spotted the al-Qaida leader, they could kill him.

This suggestion sparked a debate that, in retrospect, seems quaint. Top Air Force officers didn’t want to use Predators to kill Bin Laden because the United States wasn’t officially at war with Afghanistan or with al-Qaida. Top CIA officials were leery of the notion as well: They felt an intelligence agency shouldn’t take military action.

In January 2001, the final month of Bill Clinton’s presidency, these legal and bureaucratic issues were untangled. Top officials acknowledged the absurdity of legally permitting the U.S. government to kill Bin Laden with a submarine-launched cruise missile, as President Clinton had tried to do, but not with a smaller missile fired from a drone. And so, a modified Predator—carrying not just a camera but also a laser-seeker and a Hellfire air-to-ground missile, which could be fired by the same sort of joystick that guided the drone’s flight path—was successfully tested.

Most Air Force officers were still dubious of the new weapon, some because they doubted it would respond to signals from halfway across the world, others because the weapon’s very nature (a slow, unmanned, hovering aircraft) was anathema to the dominant Air Force culture (which cherished fast, manned jet fighters).

But the Air Force chief of staff at the time, Gen. John Jumper, was more visionary than most and backed the program. The previous May, while the debate was still raging, Jumper wrote an Air Force mission statement for a Hellfire-armed Predator, saying the weapon would be ideal for hitting “fleeting and perishable” targets. During the Cold War, this phrase would have meant Soviet armored vehicles on the plains of Europe. Now it clearly referred to cars that carried, or buildings that sheltered, al-Qaida terrorists in the mountains of Southwest Asia. In a memo to Condoleezza Rice, President-elect George W. Bush’s national security adviser, Clarke spelled out the history of the Hellfire-armed drone and urged “going forward” with new missions exploiting this new feature.

As is now well-known, Rice and the rest of Bush’s team were slow to heed Clarke and Tenet’s warnings about Bin Laden. (The first Cabinet meeting to discuss al-Qaida took place on Sept. 4, one week before the attack on the twin towers and the Pentagon.) But within the Pentagon bureaucracy, the armed drone moved ahead with unusual speed. In February 2001, three Hellfire missiles were fired from Predators at targets on a test range in Nevada. Deployment of the new weapon was scheduled for Sept. 1. Technical problems delayed shipment, but after the terrorist attacks on Sept. 11, the drones and their missiles were rushed to U.S. air bases near Afghanistan. The U.S. bombing campaign against al-Qaida commenced on Oct. 7. Five weeks later, the Predators were ready, and on Nov. 14, a Hellfire missile tore out of the seemingly empty sky and killed Mohammad Atef and six of his comrades—though it wasn’t known at the time that Atef was among those killed, and it’s still unclear, 15 years later, whether the U.S. officials running the operation even knew Atef was in the building.

A weapons system originally conceived to destroy Soviet tanks in the opening phases of World War III (even the Hellfire missile was designed as an anti-armor weapon) emerged, after a quarter-century’s gestation, as a device to kill specific people—or even a specific person—in a prolonged, global, and often secret war on terror.

Bush was keen to send more drones into action, in Afghanistan and elsewhere, but production was slow, owing in part to deliberate stalling by the Air Force chiefs. (Jumper had retired, and his successor was of a more traditional bent.) Robert Gates, who replaced Donald Rumsfeld as secretary of defense in Bush’s final two years, ordered a speed-up both in the drones’ production and delivery. By the time Barack Obama entered the White House in January 2009 (and kept Gates on for his first two years in office), drone factories were cranking out Predators and a few follow-on models nearly as fast as the U.S. commanders, mainly in Afghanistan and Iraq, were requesting them.

The new drones were a perfect fit for Obama’s philosophy of military intervention. Obama wasn’t averse to using force, but he was averse to throwing thousands of American troops into battle unless the nation’s vital interests were at stake. If the United States or its allies had a less-than-vital interest in a conflict, or if some less-than-urgent threat were posed, he preferred sending small squads of Special Forces or teams of military advisers—or, the ultimate tool of unrisky warfare, drones.

The upside of drones was that they let a president kill bad guys and ward off potential dangers without putting American troops in danger. The downside was that they made war too easy, so easy that a president, even one as smart as Obama, could convince himself that the nation wasn’t really at war—a perception belied by the people under the drone’s fire, not least the victims’ friends and family, who often started supporting or even joining America’s enemies as a result.

The reliance on drones—and the tendency to regard them as the default tool for taking out jihadi suspects on the battlefield—subsided in Obama’s second term, in part because the wars in Iraq and Afghanistan were winding down, in part because the drones weren’t having the dramatic effect that they seemed to promise. It turned out that killing a key terrorist or destroying a terrorist hangout—while sometimes fruitful and always tempting—has little impact on the course of the war. Al-Qaida had a seemingly endless line of No. 3–ranking officials to replace the ones just killed. And to the extent terrorist groups have been decimated on the battlefield, it’s been due to a combination of conventional airstrikes and forces on the ground. (When very “high-value targets” have been spotted, such as Ayman al-Zawahiri or Osama Bin Laden himself, the job is considered too important—the accuracy of a target’s ID is too vital—to leave to drones.) And sometimes, in drone strikes, innocent people get killed, not because the Hellfire missile veers off course but because the intelligence was poor, the images were fuzzy, someone has made a mistake about who was (or wasn’t) in the crosshairs. And when innocent people get killed, new terrorists—their husbands, cousins, fathers, sons, or neighbors—are often created.

The future may see more drastic blowback still. In the annals of warfare, no nation has preserved a monopoly on a new weapon for very long, and the armed drone isn’t likely to prove an exception. (It is harder to replicate than most weapons—besides the drone and the missile, you need fast data processing, communications satellites, and an airstrip near the target—but these are impediments, not brick-wall obstacles.) For now, terrorists and other enemies of America still have reason to fear the drone. It’s likely that, someday, the American officers, officials, and even presidents who decide to launch these drones will have reason to fear them too.

*Update, Sept. 14, 2016: The article has been updated to note the unsuccessful strike against Mullah Omar that preceded the successful attack that killed Mohammed Atef. (Return.)