A well researched study of video games just revealed how the background knowledge people take for granted gives us an edge over machine learning (AI).
In early 2013, DeepMind Technologies, a little-known company, published a groundbreaking result showing how a neural network could learn to play 1980s video games the way we humans do by just looking at the screen. These networks then went on to challenge and thrash the best human players.
A few months later, Google bought the company for around $400 million. DeepMind has since gone on to apply deep learning in a range of situations, most famously to outperform we humans in the ancient game of Go.
Image
It seems that we humans use a wealth of background knowledge, whenever we play a new game. And this makes the games more easier to play. But faced with games that make no use of this particular knowledge, humans flounder, whereas machines play along in almost exactly the same way.
Take a look at the computer game shown above on the left (the original game). This game is based on a classic system called the Montezuma’s Revenge, originally released for the Atari 8-bit computer back in 1984.
There is no written or in game manual and no instructions; you aren't even told which particular "sprite" you control. And you get feedback only if you can successfully finish the game.
.jpeg)
Image
By around three to five months of age,kids & infants learn to recognize object categories` by 18 to 24 months, they learn to recognize individual objects. Around this time, they also learn most properties of objects (object affordances, as psychologists call them), and they learn the difference between a walkable step along flat ground and an unwalkable step off a cliff.
It turns out that Dubey and co’s experiments ranks this kind of learned information in exactly the same order that babies (infants) learn it. “It is quite interesting to note that the order in which infants increase their knowledge matches the importance of different object priors,”
They said
“Our work takes first steps toward quantifying the importance of various priors that we humans engage in solving video games and in our understanding on how prior knowledge makes humans good at such complex tasks,” they add.
This suggests an interesting way forward for computer scientists who are working on machine intelligence—to program their charges with the same basic knowledge that we humans acquired at an early age. In this way, it is only right that machines should be able to catch up with humans in their speed of learning, and maybe even outperform them in the near future.
We will look forward to seeing the results.
That's all for today till my next post
Article Source
Thanks for Reading if this post was helpful kindly leave a vote and share your thoughts in the comments box and don't forget to Resteem so that this can help someone in need of it.
Shout out to @camzy for this toon
Don't forget to follow me @uchefrancis for more articles
My greatest fear for this AI is that, hope they won't be like Bollywood movie tittled 'ROBOT'.
Lol, I hope not
This post has received a 3.93 % upvote from @booster thanks to: @uchefrancis.
Well ladies and gents, my greatest fear is associated with the level of our own intellectual laziness and complacency. I think Mr. Eddie Izzard lays it out brilliantly in his comic relief describing our relationship with software updates. Watch it if you have a few minutes. It is hilarious but it also reveals something about our nature that is very concerning.
A healthy dose of skepticism isn't a bad thing. AI technologies are being implemented in an evergrowing range of tasks across many industries. The question should be which kind of tasks can they be really trusted with.
As many of you probably heard the prediction for the automobile industry assumes that within next 10 years all vehicles will be autonomous. I would like to recommend TED talk where Peter Haas explains The Real Reason to be Afraid of Artificial Intelligence.
It definitelly gave me a pause moment.
Aside from the topic at hand. I like how you have ended your post with a nice and slic giff animation @uchefrancis. What application have you used to make it please. If you don't mind sharing. :)