We love to marvel at the advanced robotics projects funded by the US Department of Defence because they’re so impressively — if clumsily — getting around and doing stuff like real humans. Motherboard’s latest documentary is an important reminder that today’s nerdy research will create the weapons of tomorrow.
We even joke about how Atlas is a terrifying sign we’re headed for the grim future of Terminator, but we joke because the bots are pathetic. They can barely move around, and they don’t have artificial intelligence. Calling Atlas T1000 is like joking that a baby gingerly smacking piano keys will one day be Glenn Gould.
The new doc “Inhumankind” is a sobering exploration how our robot fears might be more realistic than we think. The Motherboard crew go talk to the teams that are developing humanoid robots for the government, as well as to critics who warn about the dangers of that might arise if artificial intelligence and military technology do collide
I was taken with Nobel Peace Prize laureate Jody Williams’ stark observation that DARPA is doing R&D for the Defence Department, and that its technologies are ultimately destined to be weapons. “There is nothing the US develops in the military sphere that it does not want to weaponise,” she says in the doc. The rapid acceleration in the US military’s drone program just how quickly tech evolves from the a Defense Department hobby to an implement of war.
The bots are developed under the guise of altruism. An interview with Virginia Tech researchers developing software for Atlas reveals the best of real world intentions. They’re working on DARPA funded challenges with cash prizes, but the development is theoretically aimed at building bots that can save lives after disasters.
And sure, military robots aren’t all weapons, and yes, some save lives. Plus, dammit, developing technology is what humans do. It moves us forward. DARPA funded the research that created the internet!
Williams and the Campaign to Stop Killer Robots aren’t trying to prevent the bots from getting built, but rather ensuring they never get the autonomy to kill without human pilots pulling the trigger. The question, then, isn’t whether Atlas will be a weapon, but what kind of weapon Atlas will be. Ultimately, the problem is the future of artificial intelligence and how we choose to regulate it. When a fully evolved Atlas goes to battle, the hope is that a soldier armed with an Xbox controller is guiding the way remotely.
Watching the six-six hulk lurch around is evidence enough that this tech is still a long way from nightmare sc-fi bots. But if like us at Giz, you’re amused every time Atlas takes a step, just remember you’re watching the baby steps of a future killer. [Motherboard]