AI Knows From A To Z

 

 

Target found out the pregnancy of a teenager before her parents did.

An angry father walks into a Target store in Minneapolis, demanding to talk to the manager:

“My daughter got this in the mail!” he said. “She’s still in high school, and you’re sending her coupons for baby clothes and cribs? Are you trying to encourage her to get pregnant?”

A few days later:

“I had a talk with my daughter,” he said. “It turns out there’s been some activities in my house I haven’t been completely aware of. She’s due in August. I owe you an apology.”

Target had a system which assigns each shopper a “pregnancy prediction” score based on the products they buy. The system could also estimate their due date to within a small window, so Target could send coupons timed to very specific stages of their pregnancy.

This happened in 2012 and it’s hardly state-of-the-art “AI”, but it just goes to show that anything creepy a machine learning model does, is just a product of how and with what data it is trained.

 

Programmer and CMU PhD Tom Murphy created a function to “beat” NES games by watching the score. How it worked was that the program would do things that increased the score, and then learn how to reproduce them again and again, resulting in high scores. It came up with novel techniques and strategies for playing games and even exploited glitches humans didn’t know about, or at least hadn’t told it about. The program, called a “technique for automating NES games,” can take on nearly every NES game. Nearly.

Tom made the program play Tetris. Most of us have played this game and needless to say, we all know that it gets tricky after a certain point. The program struggled to figure out what to do. The choices of Tetris blocks is entirely random, so it’s not surprising that the computer wasn’t able to consider future repercussions far enough ahead to notice that stacking those blocks in a certain ways made a big difference.

On one such run, when faced with imminent defeat, the computer did something eerie. Rather than lose, and receive a ‘game over’ message, it just paused the game. Forever.

Tom describes the computer’s reasoning like this: “The only winning move is to not play.” And that’s right. If you pause a game for ever you will never lose that game.

 

An Artificial Intelligence program developed by Elon Musk’s Team called Open AI created a lot of buzz as well. Musk believes that development of AI should be regulated and AI safety should be a prime concern of every developer. To put weight to his idea, he started a project called OpenAI. The team used DOTA 2 as a test means to develop their AI.

Now what’s special is how they trained this BOT. They didn’t write any code about the rules of Dota 2 or any strategies that professional players use. They just gave basic instructions(eg: Winning is good, losing is bad, Taking Damage is Bad, Giving Damage is good, etc) and made the BOT play with a copy of itself. In the beginning, the BOT made very stupid decisions. But slowly, it started to learn, devise its own strategies and make novel moves. It took the BOT 2 hours to beat the existing Dota 2 BOT and 2 weeks to reach the level of a professional Dota player!

Finally, OpenAI put its BOT to test against many of the world’s top Dota 2 players(1v1 match) and it was easily able defeat them. Then came The International 2017, one of the biggest eSports event in the world. Here, OpenAI was tested against what people consider the best Dota 2 player in the world: Danylo “Dendi” Ishutin. To everyone’s surprise, OpenAI defeated Dendi in a solid 2–0 before Dendi gave up!

 

A Few More Worthy Mentions 

 

The blink recognition software in Nikon’s camera kept asking “Did someone blink” when an Asian would pose in front of the camera. The camera perceived the *small* eyes of Asians as closed.

Recently a report announced that Facebook had to abandon their experiment after two AIs went out of control and supposedly started interacting with each other in a language other than English which made them easier to work! below is what they said to each other.

Bob: i can i i everything else . . . . . . . . . . . . . .

Alice: balls have zero to me to me to me to me to me to me to me to me to

Bob: you i everything else . . . . . . . . . . . . . .

Alice: balls have a ball to me to me to me to me to me to me to me

Bob: i i can i i i everything else . . . . . . . . . . . . . .

Alice: balls have a ball to me to me to me to me to me to me to me

Bob: i . . . . . . . . . . . . . . . . . . .

Alice: balls have zero to me to me to me to me to me to me to me to me to

Bob: you i i i i i everything else . . . . . . . . . . . . . .

Alice: balls have 0 to me to me to me to me to me to me to me to me to

Bob: you i i i everything else . . . . . . . . . . . . . .

Alice: balls have zero to me to me to me to me to me to me to me to me to

You, i everything else (dot)(dot)(dot)(dot), looks like Bob was devising a plan to kill everyone other than him and Alice.

 

And that ladies & gentleman is how Artificial Intelligence has evolved over the years.

Fun fact, it always is on the move. It is always evolving.