For many years I have criticized that video "Slaughterbots" about the autonomous assassin quadcopter drones with facial-recognition tech. And I still do, it was silly and alarmist. But some elements of it sure seem to be coming close to reality at this point.
Not yet, there are several key parts of the "slaughterbot" scenario's technology that's still missing. In the video's scenario the quadcopters had real-time facial recognition and autonomous navigation that allowed them to fly around inside a university looking for specific individuals to kill in a chaotic swarm of panicked people, which is still well beyond anything we've got right now. Especially in terms of cheapness and reliability. And also as I recall the film skipped over the fact that quadcopters have no way of opening doors. Even in this real-world scenario a simple curtain would have prevented the drone in this video from getting inside the building to detonate, which would have reduced its deadliness significantly.
Maybe someday. But by then I would expect there to be autonomous defender bots zooming around dog-fighting enemy slaughterbots and things will balance out again between offense and defense.
Given that the drones wouldn't know which door they need to open, this sounds like it'll get rapidly very expensive and not so much of a targeted thing any more.
I see it the other way around. Virtually all modern mobile phones can easily do real time facial recognition, and certainly have enough processing power to navigate through a known location. And even being cynical about tracking specific individuals, targeting all humans would be an easy problem to solve in comparison, assuming you’re just trying to sow chaos.
As for the problem of doors, in the film there were specialized drones with shaped charges to blow holes in walls and windows, letting the rest of the swarm inside.