Slashdot Mirror
Why Humans Learn Faster Than AI (technologyreview.com)
What is it about human learning that allows us to perform so well with relatively little experience? MIT Technology Review: Today we get an answer of sorts thanks to the work of Rachit Dubey and colleagues at the University of California, Berkeley. They have studied the way humans interact with video games to find out what kind of prior knowledge we rely on to make sense of them. It turns out that humans use a wealth of background knowledge whenever we take on a new game. And this makes the games significantly easier to play. But faced with games that make no use of this knowledge, humans flounder, whereas machines plod along in exactly the same way. Take a look at the computer game shown here. This game is based on a classic called Montezuma's Revenge, originally released for the Atari 8-bit computer in 1984. There is no manual and no instructions; you aren't even told which "sprite" you control. And you get feedback only if you successfully finish the game.
Would you be able to do so? How long would it take? You can try it at this website. In all likelihood, the game will take you about a minute, and in the process you'll probably make about 3,000 keyboard actions. That's what Dubey and co found when they gave the game to 40 workers from Amazon's crowdsourcing site Mechanical Turk, who were offered $1 to finish it. "This is not overly surprising as one could easily guess that the game's goal is to move the robot sprite towards the princess by stepping on the brick-like objects and using ladders to reach the higher platforms while avoiding the angry pink and the fire objects," the researchers say. By contrast, the game is hard for machines: many standard deep-learning algorithms couldn't solve it at all, because there is no way for an algorithm to evaluate progress inside the game when feedback comes only from finishing.
Would you be able to do so? How long would it take? You can try it at this website. In all likelihood, the game will take you about a minute, and in the process you'll probably make about 3,000 keyboard actions. That's what Dubey and co found when they gave the game to 40 workers from Amazon's crowdsourcing site Mechanical Turk, who were offered $1 to finish it. "This is not overly surprising as one could easily guess that the game's goal is to move the robot sprite towards the princess by stepping on the brick-like objects and using ladders to reach the higher platforms while avoiding the angry pink and the fire objects," the researchers say. By contrast, the game is hard for machines: many standard deep-learning algorithms couldn't solve it at all, because there is no way for an algorithm to evaluate progress inside the game when feedback comes only from finishing.
With trump as president, one could argue that humans never learn at all.
We haven't developed anything resembling actual AI yet. These systems are simple brute force machine learning or "deep" learning systems. Nothing fancy or special about them. They are a tool and nothing else. Any decision they make has already been pre-planned by their human programmers. They are not in fact learning from something and applying that to something else.
While they might be taught to open a jar of peanut butter, they would become confused if you presented them with a screw top bottle of wine and wouldn't be able to open it. However, a 2 year old, if they had the dexterity, once taught how to open a jar of peanut butter would then apply that same knowledge to any other jar and not have to be shown millions of different jars. These machine learning systems cannot yet think in generic terms and be shown something once and apply that to multitudes of different applications. Look at the image posted the other day about a machine learning system incorrectly called "AI" confusing rocks with sheep. Not even someone with an IQ of 60 would have made that mistake.
When we do finally start to crack the actual AI egg in a 50-100 years, then those system will start learning something instead of having to be told (millions or billions) of times what something is.
I already finished.
Sounds like what LSL would say to his "date" for the night.
https://app.box.com/WitthoftResume Code: https://github.com/cellocgw
IQ tests and the like suffer from cultural bias. So do these games. As an English speaking human who has been exposed to pop culture and movies, I know that movement from left-right is normal, the hero saves the princess (which is a problem and another discussion). I wasn't born with this bias. I learned it. It is silly to initialize a neural networks with random weights (as if it was just born) and then declare it learns slower than a human. Let a computer create a game where a normal human cultural bias don't apply and have an infant play. Then we will see a more accurate comparison.
This is like handing a chess set to an isolated Amazon tribe and only tell them "sorry, invalid game" until they make a valid checkmate. They'd probably never even find the opening position, much less make any correct moves and certainly not how to mate. They'd just randomly do things until they got bored or made up their own game. There's no reason a machine should expect "getting to the top" to be a valid objective without a whole lot of insight into the human condition and "because it's there".
Live today, because you never know what tomorrow brings
Humans have also learned how to learn. Whether we realize it or not, we learn something every day of our lives. We are used to learning and since we practice it continually, we're good at it. Our survival depends on it.
I'm 40 years old. I spent the first 6 years of my life figuring out the world around me, the years from there to about 18 learning stuff in school and figuring out how to use complicated machines and understanding the deeper rules of society and complex systems (like how the rules surrounding driving work; not just the legal rules but also the implicit social conventions), and I've spent the 22 years since then refining my understanding of the world and my place in society. I program computers (and games!) for a living and study philosophy and ecology as hobbies.
Why can I figure that game out faster than a newly hatched AI? Literally everything in my life over 40 years led up to the moment when I played that game. It's really not a fair fight.
So... your bottle of anti-psychotics are talking to you? Maybe you need a different prescription.
Starships were meant to fly, Hands up and touch the sky - Nicky Minaj
1. It's still obviously a platform game, even at the 'hardest' level. Try it and tell yourself platforming experience doesn't matter. You can even sort of recognize the ladders (which, given the goal of the game, are pretty crucial).
2. The algorithm they are comparing against is designed for exploration, not for getting to a goal as quickly as possible. See: https://pathak22.github.io/nor...
Note also that that comparison was not about pitting humans against the best algorithm for this specific game, but to highlight the point that there is information in the specific representation in objects that humans use and NN algorithms don't (yet).
The point of the study is effectively to refute the first line of the summary:
"What is it about human learning that allows us to perform so well with relatively little experience?"
Answer: we don't perform well in those circumstances. We all have many years, or decades, of experience to draw on, and we do.
The fact that AlphaGo Zero taught itself how to play Go much better than the thousands of years humans have had to perfect the game shows AI is capable of learning much faster than humans.
At this point, AI naysayers will go for the goalpost shifting argument tactic.
The more I talk to people, even the very intelligent, the more I see that humans don't really learn all that well. People tend to literally refuse to learn.
Those who do not learn from commit history are doomed to regress it.