The Twitterverse panicked briefly last week after an insanely scary video of a robot turning on its creators was circulated and trended. Turns out, it wasn’t real. The original parody video (posted above) had been spoofed and circulated after being edited further, with the intention of scaring the world into thinking that the robot revolution is now happening and we’re now living in the future work world of our darkest fears. The one where hostile robots exact revenge on their humans.
In the full original clip of the video, an earnest and hardworking robot is tormented and harassed by its creators while he’s simply trying to do his job. Once the bot has had enough, it turns on its keepers in full-on Terminator super-bot fashion. Watch through to the end and you’ll see behind-the-scenes video that shows it’s clearly a parody, and that the robot is actually CGI.
Late last week, however, Twitter user @kocizm posted a trimmed, low-quality version of the parody video that edited out the clarifying parody part at the end. The edited video has since been removed for copyright issues, but not in time to stop more than 100,000 retweets and comments indicating that people were pretty freaked out.
So what’s the lesson in all of this?
Perhaps there are a couple. First, it serves as a reminder to never immediately believe that everything on the internet is real. In the bigger picture, it is genuinely disturbing to see how the “creators” bully the robot in the parody video, like mean bosses or coworkers. It’s hard not to feel sorry for the robot.
Even though the viral version of the video — in which the armed robot takes revenge — is a spoof, it could serve as a reminder that as robots continue to join humans the workplace, it’s a really good idea to be nice or, at the very least, not mean to them, and maybe not because of the deep dark fear that they will turn on humans Terminator-style. As Michael Schrage writes for Harvard Business Review in his article on Why You Shouldn’t Swear at Siri, as digital devices grow smarter, being beastly toward bots could cost you your job.
In order for the technology to work, robot and AI interfaces record and synthesize everything that they’re asked to do, and they often make mistakes understanding voice commands until, after much trial and error, the algorithm gets it right. About 10 percent to 50 percent of interactions are abusive, according to computer scientist Dr. Sheryl Brahnam. These “recordings,” or in a professional setting “logs,” of employees cursing at or even breaking and abusing robots and workplace AI interfaces could easily be played back to an HR person. And, perhaps, someday that could limit an employer’s view of someone’s ability to work with technology.
A real-life example of this, outside of the workplace, played out when Microsoft’s Tay became a real-world case study of how abusive language shapes AI and, in this case, a chatbot’s responses. Less than a day after Microsoft research released the Twitter bot unsupervised, the chatbot from hell was tweeting a firestorm of inappropriate, dirty, racist, and homophobic comments until Microsoft pulled Tay’s plug. But the chatbot was simply doing what it was programmed to do by responding in the way it learned to respond based on the way its human users were talking to it.
“These behaviors are simply not sustainable. If adaptive bots learn from every meaningful human interaction they have, then mistreatment and abuse become technological toxins. Bad behavior can poison bot behavior,” writes Schrage. “That undermines enterprise efficiency, productivity, and culture.”
We like Schrage’s suggestion that “being bad to bots will become professionally and socially taboo in tomorrow’s workplace. When deep-learning devices emotionally resonate with their users, mistreating them feels less like breaking one’s mobile phone than kicking a kitten. The former earns a reprimand; the latter gets you fired.”
Like what you read? Check out more from my WorkingNation blog, The Looming Robot.