The danger of AI is weirder than you think

The danger of AI is weirder than you think 

The danger of AI is weirder than you think 

Artificial intelligence

Artificial intelligence is known to hinder the disruption of all types of industries, and what about ice cream, any kind of new flavor that the mind can generate with the power of advanced artificial intelligence, so I collaborated with a group of programmers from Keeling Middle School to discover the answer to this question and collected more than 1,600 Flavor there is ice cream, and together we fed it to an algorithm to find out what it would generate, here are some flavors that Amnesty International came up with these flavors are not delicious and because we were hoping this would be so, the question is: what happened, did Amnesty International try to kill us or is it T was trying. He did what we asked for and there was a problem with the movies when an AI error occurred.

Artificial Intelligence decided that it no longer wanted to obey humans

Artificial Intelligence has decided that it no longer wants to obey humans, and has its own goals, thank you very much in real life, although the AI ​​we already have is not smart enough for that as an approximate computing power of Earthworm or perhaps at least one bee, perhaps actually Maybe less like we are constantly learning new things about minds that show how AIS doesn't measure our real minds, so today AI can do things like defining pedestrians in a picture but it has no concept that goes beyond pedestrians is a whole lot of lines, tissues and unknown things that actually exist are that That, so we're going to AI today what we asked to do if he could, but he might not do what we really want, let's say you're trying to get AI to take this set of robot parts and assemble them into some kind of robot going from point A to point B now if I was going to try to solve this problem by writing the traditional style progr I computer I would provide step-by-step instructions for the program on how to take these parts in assembly mode and a legged robot then how to use these legs to denote point B, but when artificial intelligence is used to solve problem E, Different from how to solve the problem, you are given a goal For alone,

She should introduce herself through experience about how to reach this goal and demonstrate that the way in general Amnesty International will deal with this specific issue as such.

It collects itself in a tower and then falls and falls at point B, and this technology solves the problem. You have reached point B if Amnesty International does not rebel against us if this will exactly lead to what we are asking for, then a hoax ... How do we put the problem in working with Amnesty International until we do what we really want to do? This little robot is condemned by AI, and

AI designed robot legs 

AI designed the legs of the robot and then discovered how to use them to overcome all these obstacles, but when David Haass created this experiment, he had to prepare it with very strict restrictions on the large size that allowed AI to do if it were otherwise, and traditionally, You've reached the end of this hurdle, of course, so you can see how hard it is to get AI to do something as simple as walking, so seeing AI does that is okay. It is fair that you cannot just be a tall tower and you really have to like to use legs for walking, and it turns out that this does not always work well, so this eye function was moving fast and didn't work, tell her that she had to run or that he couldn't use his arms, This is what you get when you train AI to move quickly,

You can get things like somersaulting, and Silly Walks is completely normal, so shudder to the ground in a pile, from your assessment that you realize it should be the weirdest thing a scraper robot does to jump forward which is another thing AI does if you I gave him a possibility

Artificial Intelligence Training in Simulation 

If you train artificial intelligence in simulation, you will learn how to do things like penetration through simulated math errors and harvest them for energy or will determine how to move faster by repeating over and over again on the ground when working with artificial intelligence like working with another human that is similar to working to the point Large with a kind of bizarre nature of nature, and it's really easy to give artificial intelligence the wrong problem to solve, and often we don't realize it until something goes wrong