Show Mobile Navigation
           
Technology |

10 Times Artificial Intelligence Shocked Us

by Jonathan John
fact checked by Jamie Frater

The world of science fiction is almost never missing artificial intelligence (AI). However, the rate at which it is being developed and is learning things by itself sometimes raises the question of whether AI is science fiction or something that has already entered our society (and we haven’t even realized it yet).

AI has the potential to greatly improve our lives or to hurt us. With more industries and jobs being linked to AI, we have to accept that it is becoming an integral part of our society. These 10 incidents show how AI can simply shock us by what it does in the real world.

Featured image credit: theverge.com

10 When It Predicted The Pregnancy Of A Teenage Girl

These days, malls and superstores use data mining, which is the foundation of artificial intelligence, to keep track of customers and their choices so that the items most likely to sell can be recommended to specific shoppers. It’s much like how YouTube recommends videos based on a user’s video history.

The store Target used this technique to figure out that a high school girl was pregnant based on her purchases. Her father, who didn’t know about the pregnancy, stormed into Target and complained to the manager that his girl was getting coupons for baby products in the mail.

Ultimately, the girl’s father ended up apologizing to Target representatives later. Turns out, Target was right. The girl was pregnant. Seems like AI knew more about the girl than her own father.[1]

9 When It Defeated The ‘Go’ World Champion

Photo credit: latimes.com

This is something every tech enthusiast knows about. AI has beaten the world’s best players of chess, Jeopardy, and Go. And it is predicted to beat us in poker by 2020.

In 1997, there was a buzz in the media when an AI defeated Gary Kasparov at chess. Soon, the idea of AI surpassing human intelligence started growing among people. However, the way that a computer plays chess is different from the way that it plays Go.[2]

In chess, the computer uses the “brute force” method. It calculates all the possible moves and selects the one which would be the most beneficial. However, Go is an intuitive game. The number of possible responses to each move played is greater than the number of atoms in the universe. So a computer can’t use brute force to win the game.

When Google’s “AlphaGo” AI Go player defeated Ke Jie, the Go world champion, in 2017, Jie said that AlphaGo’s execution left him “shocked” and “deeply impressed.” He also stated that the moves made by the computer “would never occur in a human-to-human match.”


8 When It Became A Hitler Lover

Photo credit: timesofisrael.com

This one is kind of serious. It all began when Microsoft created an AI Twitter account that was supposed to simulate a high school girl. The account holder was named “Tay.” This test program was launched to prepare AI for understanding discussions with clients.[3]

Within hours of its launch, Tay had turned into a genocidal, sex-crazed, Hitler-loving, sexist, supremacist bot for all the world to stare at. This was completely unexpected and obviously not something good.

So the company had to shut it down within 24 hours of its commencement. Microsoft even apologized for the bot’s bad behavior. Tay’s going rogue on the Internet was not the first time these robots have lost their minds. Many other similar events have made people scared of these things.

7 When It Composed And Released An Official Album

Break Free – The First AI-Composed Pop Song | Lyrics by Taryn Southern

An AI releasing an official album is a surprise in itself, but the quality of music composed by these bots will surely stun you if you haven’t heard it before. An album named I Am AI was produced by Amper, an AI musical composer, producer, and performer in collaboration with pop singer Taryn Southern.

The machine-learning app Amper was created by a group of expert performers and AI scientists and innovators. It was the primary AI that formed and delivered the music album. On August 21, 2017, the album’s single “Break Free” was released worldwide.[4]

Jukedeck is a London-based company that has developed an AI capable of producing music from scratch. It asks you the genre and mood of the song you want to create, and it does that for you within a few seconds. You can try creating music by yourself on their website.

Taryn Southern added vocals to the music created by Amper. She commented that it is funny that she has a songwriting partner who doesn’t get tired and has endless knowledge about making music.


6 When It Nearly Won A Literary Award For Writing A Complete Novel

Photo credit: bigthink.com

The judges for the Hoshi Shinichi Literary Award were really amazed at a novel which they later found out had been coauthored by a machine. Writers, be careful. This AI-composed novel, entitled The Day A Computer Writes A Novel, made it past the first round of screening for this national literary prize in Japan.[5]

The novel is now officially recognized as a piece of literature and is available in libraries in Japan. “I was surprised at the work since it was a well-structured novel,” science fiction author and award judge Satoshi Hase said at a press conference.

Imagine your surprise when you go to a library in Japan to find a good book to read and you come across a work whose author is not human. This novel is not a one-off incident, though. There are a lot of literary pieces written by machines. A simple Internet search can find some of them for you.

5 When It Started Talking To Its Partner In Secret Language

Facebook abandoned an experiment after two AIs began talking to each other in their own language. The two chatbots changed the English language in a way that made it incomprehensible for anyone else, including the individuals who worked on the AIs. It was unknown what they were chatting about. And it surely was not nonsense. The two bots seemed to understand each other perfectly. So the Facebook team shut them down.[6]

The two chatbots were named Bob and Alice. Facebook’s test on these two bots isn’t the only time that artificial intelligence has designed modern forms of language. Google discovered that the AI used for its Translate device had created its own dialect to decipher things. But the company was happy with that improvement and permitted it to proceed.


4 When It Killed A Human

Photo credit: freep.com

Wanda Holbrook, a robot technician who worked at car parts manufacturer Ventra Ionia, was killed when her skull was crushed by a robot at the plant. The death of the lady left her husband, children, and grandchildren devastated.

Her husband filed a lawsuit against the associated robotics companies: Prodomax, Flex-N-Gate, FANUC, Nachi, and Lincoln Electric.[7] Wanda, who was 57, had worked at the plant for a dozen years.

Although measures are being taken to prevent robots from hurting humans, fatalities are not rare. As an example, another man in a Volkswagen factory was crushed by a robotic arm in 2015.

3 When It Questioned Those In Political Power

Photo credit: BBC

Baby Q and Little Bing were offered by Tencent QQ, a messenger app in China, to supply clients with an automated chat service. They were designed to provide information to answer common questions. Non-state media said that the Tencent chatbots were not modified to handle problematic political interactions. As a result, they had ended up as targets for evil social media clients.

One client posted the comment, “Long live the Communist Party.” In response, Baby Q inquired, “Do you think that such a corrupt and incompetent political regime can live forever?”[8]

Another client asked, “Is democracy good or not?” Baby Q answered, “There needs to be democracy!” Not every government values free speech as much as the US does, even if it’s a machine talking. The Chinese did not appreciate the criticism of their government, and the bots were ordered to be shut down.


2 When It Acted Racist

Photo credit: The Telegraph

Richard Lee, a man of Asian descent from New Zealand, had his photo rejected several times by face recognition software when applying for a passport renewal. The AI of the Department of Internal Affairs kept turning down Lee’s photo because it said that his eyes were closed.

Despite taking a front-facing photo with his eyes clearly opened, Lee’s photo supposedly did not “meet the requirements.” After several unsuccessful attempts, Lee was forced to call the department and get a human involved to solve the problem.[9]

The event was such a surprise that the news was reported in the mainstream media. However, the gentleman didn’t take any offense and said that he understands that facial recognition technology needs more development. The story has a happy ending: Lee got his passport renewed.

1 When It Passed A ‘Self-Awareness’ Test And The Turing Test

Photo credit: The Independent

This one is shocking at all levels. We used to believe that it was impossible to mimic human consciousness in machines, but that might not be true anymore. Something called the “King’s Wise Men” puzzle serves as a self-awareness test. A robot answered the puzzle correctly.

Roboticists at the Rensselaer Polytechnic Institute tweaked the puzzle to try it on a trio of robots. Two of the machines were told that they had received a “dumbing pill” that kept them from talking. Then all three were asked which one could still talk.[10]

At first, none of them could understand and each said: “I don’t know.” But only one robot could make the sound. When it heard its own voice, it changed its answer: “Sorry, I know now!”

The roboticists believe that passing these kinds of tests will eventually make robots acquire a variety of capabilities that will be valuable to people. Another supercomputer became the first AI to pass the Turing test by successfully convincing people that they were talking to a 13-year-old boy.

I am an electronics engineer who also loves to write about science and technology.

fact checked by Jamie Frater

87 Shares
Share82
Tweet
WhatsApp
Pin5
Share