We knew the Boston Dynamics robots could dance, but this is something else. Now, the company's much loved Atlas robot has learned to bust some gnarly parkour moves, backflips and all (above), but there's a pretty sensational blooper reel to go along with it (further down).
I'm not sure whether to laugh or cry right now, but these outtakes of Atlas falling flat on its arse (via Futurism) are sure to awaken some flavour of emotion for you—be it joyous or fearful.
It took a month of training to turn Atlas into the hardcore parkour star it is today. Just as a babe learns to walk, its newfound proficiency was underpinned by countless tumbles. Each fall helps the AI build a more concrete understanding of the course and how best to conquer it.
After all that hard work and dedication, it's time to sit back and reap the benefits. Among which sits this incredible outtakes video, cut from a longer video on the Boston Dynamics Youtube channel, and tweeted by science communicator Kyle Hill. Enjoy.
Reminder that science is still hard pic.twitter.com/1leibXqXBLAugust 17, 2021
Best CPU for gaming: the top chips from Intel and AMD
Best graphics card: your perfect pixel-pusher awaits
Best SSD for gaming: get into the game ahead of the rest
Earlier this month, we saw a new breed of AI pretty much ace the Blade Runner Voight-Kampff empathy test, and Atlas doing parkour really takes the biscuit for me. It's just another example of how quickly AI is adapting to the strange human-made world it's confronted with. Hiccups are to be expected, of course, as with any tech project—not least one that involves highly mobile artificial intelligence.
So, AI takes another leap toward infiltrating the fabric of our everyday lives. The latest advancements mean AI can be your conversational partner, dance partner, and now even parkour partner. And you know Atlas will end up better than us.
With Elon Musk's plans for the Tesla bot surfacing, it seems the age of artificially intelligent butlers will soon be upon us. But the bloopers here show there's still a long way to go before we get them right.