A big AI upset in telecom is growing scarily possible

Clad in a Zootopia apron emblazoned with "I'm good," the dance bot had seemingly rebelled, banging its fiberglass fists on a customer's table. Before food and drink erupted, a semi-amused waitress tried to drag it away by the collar with its arms still swinging, as hard to restrain as a toddler mid-tantrum. An off button was either inaccessible or simply didn't exist. This initial glimpse of rogue physical AI was filmed at a restaurant in California and has predictably gone viral. Developers had evidently not considered a scenario in which the robot is moved too close to a customer's table and had not built in a failsafe. As humanoids multiply, this is all a bit scary. Bad network software cannot, at least, cause direct physical harm. An accidental clout by a four-limbed steel acrobat might sting. Isaac Asimov, were he alive, would undoubtedly be shaking his head and muttering angrily about his three laws of robotics. Inevitably, robots will soon be unleashed on humans in countries where any AI regulation now seems to be viewed by governments as a sign of technological timidity. Before then, a non-physical AI "accident" will happen. It may have physical consequences, and it could well be in telecommunications. It could just as easily be mundane, quickly fixable, and limited in scope. Some would argue that numerous small accidents have already occurred.