8 Hilarious Mistakes Made by Artificial Intelligence

We may rely on our friendly robot or AI system to help us humans out from time to time. Here are eight funny times they weren't so helpful.

Real robots are taking over the world.

Sure, you’ve heard that before. You even remember the Twilight Zone episode that warned us about it 60 years ago. But now it’s getting real. One android recently published a novel. At Café X, in San Francisco, robot baristas make and serve coffee, and another California restaurant chain, Caliburger, is trying out a robot that can flip 2,000 burgers a day. What human can compete with that? Especially when you consider that androids don’t complain, ask for raises, or microwave fish in the break room.

But before we hand over the keys of society to our automaton overlords, let’s recognize that there’s another side to robot-kind—one that’s all too human. To celebrate the 100th anniversary of the coining of the word robot, by the Czech playwright Karel Čapek, we thought it would be fun to take a look at a by-no-means-complete list of AI’s failed attempts to replace us flesh-and-bones types.

Hold the Beothurtreed!

Janelle Shane, an optics research scientist, wanted to find out if artificial intelligence could create a menu that didn’t taste, well, artificial. So, she fed a computer 30,000 cookbook recipes, then programmed it to create its own recipes. The result? Let’s just say McDonald’s has nothing to worry about. One recipe the computer created was something called “Beothurtreed Tuna Pie.” Want to make it? You’ll need the following ingredients:

  • “1 hard cooked apple mayonnaise”
  • “5 cup lumps; thinly sliced”
  • “Surround with 1½ dozen heavy water by high, and drain & cut in ¼ in. remaining the skillet”

Of course, that’s how everyone’s grandmother cooked Beothurtreed Tuna Pie. The computer’s other specialty is the “Tart Cover Shrimp Butter Wol,” which calls for something we all have sitting in our pantry: “1 can fried pale fruit to cover the drain.” Down the drain is probably where the meal will end up after one bite. So will another dish, which required “1 cup grated white rice.” There’s a lot more where this came from, but you get the idea. Turn down any invitation for dinner this computer may send you.

What service!

A few years back, the Henn-Na hotel in Nagasaki, Japan, hired 243 robots to cover duties ranging from concierge to bellhop. But not long after the experiment began, it ended with managers “firing” half of the robots because they kept malfunctioning. The check-in robots had trouble answering guest’s questions and photocopying passports, while bellhop robots kept banging into walls and tripping over curbs. If a guest wanted to sleep in late, too bad. One in-room assistant kept waking up a lodger every time he snored, saying, “Sorry, I couldn’t catch that. Could you repeat your request?”

You get a dollhouse! And you get a dollhouse!

We’ve all heard of humorous incidents where kids have used their friendly Amazon Echo or Google Home to order up whatever their heart desired. Another such run-of-the-mill incident occurred in Dallas, when a six-year-old ordered a dollhouse (and four pounds of cookies, for a snack!). What was not run-of-the-mill was what happened a few days later when the local news reported the whimsical moment. When the news anchor stated, “I love the little girl saying, ‘Alexa ordered me a dollhouse,'” many viewers (and Echo owners) found that their own Echo devices had set about ordering dollhouses for them.

Stop the presses!

News outlets are turning to artificial intelligence to create content, including weather and quarterly earnings reports, as well as sports recaps—anything data driven that doesn’t necessitate a human’s touch. But that doesn’t mean the results will be any better. In 2017, the Los Angeles Times published a story about a 6.8 earthquake that shook Santa Barbara, California. You would expect such a large quake to have gotten a lot of press coverage. And it did…in 1925, when the earthquake happened. Turns out the report was produced by a computer program called the Quakebot, which generates articles based on notices from the U.S. Geological Survey (USGS). When a staffer at the USGS was updating the historical data, the QuakeBot got a little confused.

You look familiar

Facial recognition can also prove confusing. Two years ago, Amazon’s vaunted facial recognition software matched 28 members of Congress to mugshots of criminals. Oops. The Scottish soccer team Inverness Caledonian Thistle FC bypassed facial recognition in favor of ball recognition by replacing their human camera operators with AI-operated ball-tracking cameras. Now, cameras would always follow the action by automatically following the ball. Sounds great, except fans watching at home missed most of the scoring plays as the AI-operated cameras constantly mistook the referee’s bald head for the soccer ball. Scores of viewers called the team to complain, one going so far as to suggest supplying the ref with a toupee.

robot incorrectly cooking an egg directly on a stove burner with it still in the shellRichard Borge for Reader's Digest

Party for one

One night in Hamburg, Germany, one Amazon Alexa took the evening’s entertainment into its own, er, hands. At about 1:50 a.m., this particular Alexa started playing music at such exceedingly high volume that neighbors were forced to call the police. The police arrived and knocked, but, of course, no one was there to answer the door. They broke in and pulled the plug on the device. As a parting gift, they left a new lock on the door, for the homeowner to discover when he returned. He was forced to head to the police station to pick up the new key—and pay the considerable locksmith bill. He and his Alexa have since parted ways, after this “U-turn” in their relationship. If you and your Alexa are still on good terms, here’s how you can prolong your device’s life.

Quick, get me a broom!

What’s the worst a robotic vacuum cleaner can do, right? Let Jesse Newton count the ways for you. Newton took to the Web to share that his new puppy had pooped on the floor at 1:30 in the morning while he and his wife were asleep. How did he know the exact time? “Our Roomba runs at 1:30 a.m. every night,” he wrote. “And it found the poop. And so began the Pooptastrophe. The Poohpocalypse. The Pooppening.” The robot vacuum spread the puppy’s load throughout the house, decorating floorboards, furniture legs, and rugs, “resulting in a home that closely resembles a Jackson Pollock poop painting.” Another potentially scary thing? Here’s what your robot vacuum might know about your house.

Whatever you do, don’t upset Sophia!

It’s pretty clear that robots are not perfect. Still, they’re not going anywhere. After all, they’re here to serve us, right? Take Sophia, a social humanoid robot developed by Hanson Robotics. She/it has the face of an attractive woman and the ability to hold a conversation, much like Apple’s Siri, making her/it eerily human-like. When CEO David Hanson and Sophia appeared on CNBC’s The Pulse, he himself asked the AI what was clearly on the mind of many people in that studio: “Sophia, do you want to destroy humans?” Without hesitation, Sophia—smiling a tad too broadly for our taste—responded, “OK, I will destroy humans.”

Reader's Digest
Originally Published in Reader's Digest