Artificial Intelligence and Machine Learning is amuch-discussed topic in and outside of the tech community these days. In the coming years, we could likely see Artificial Intelligence being used in almost every horizon right from occupying the spaces of your home to large and gigantic industrial applications.

So to explain in laymans terms, the A.I of Machine Learning technology aims at making self-learning robots that in future can exhibit human-likecapabilities and perform similar tasks.

Many tech giants like Facebook, Google, Amazon, Tesla and others have already started working in this segment. However, there are some people expressing concern about developing AI because of the potential threat to the existence of humans.

Stephen Hawking has previously said that AI will either bethe best, or the worst thing, ever to happen to humanity.

Similarly, Tesla boss and CEO Elon Musk, while talking about AI, saidAI systems today have impressive but narrow capabilities.It seems that well keep whittling away at their constraints, and in the extreme case, they will reach human performance on virtually every intellectual task. Its hard to fathom how much human-level AI could benefit society, and its equally hard to imagine how much it could damage society if built or used incorrectly.

Some interesting results have recently come out of the tests performed with Googles DeepMind AI system. In this test, the Google AI demonstrated an ability of self-learning from its own memory.

The team at DeepMind AI ran about 40 million tests for a fruit gathering computer game wherein two DeepMind agents have to compete to gather the maximum number of virtual Apples.

Initially, with enough Apples to gather, things went well between the two agents. But as soon as the number of available apples began to reduce, the two agents became extremely aggressive to the point of using laser beams to defeat each other down to gather more Apples.

Take a look at the below video:

Studying this behavioral pattern, the researchers at Google said that DeepMind feels that it is about to lose, and hence it turns aggressive to come out in the top spot.

The algorithm was designed like if one agent tags the opponent with a laser beam, it would leave the opponent out of the game for some time. Meaning the first one could then gather more Apples and ultimately win.

Rhett Jones from Gizmodo, tells that for simpler algorithms DeepMind used smaller agents and demonstrated peaceful co-existence among themselves.

However, for complex networks, the agents turned competitive and aggressive, each willing to defeat the other one.

However, the Google team performed another test with three AI agents in a video game called Wolfpack. Two of the three agents are wolves while the third is a prey.

Unlike the aggressive behavior inGathering,the two wolf agents exhibited a cooperative behavior. This is because the algorithm would reward both the wolf agents with extra points once they come near the prey, regardless of who takes away the prey.

In their paper, the Google team writes The idea is that the prey is dangerous a lone wolf can overcome it, but is at risk of losing the carcasses to scavengers.However, when the two wolves capture the prey together, they can better protect the carcass from scavengers, and hence receive a higher reward.

We are just in the nascent stage of development for Artificial Intelligence, and these results successfully exhibit and early behavior of humanness filled with competitive nature.

This technology is likely to have a lot of merits and make multiple tasks absolutely easy for humans. But we believe that developers should take every care in their creations.

For all of you who are interested in knowing more about it and analyze the data of these experiments, you can have a look at the official DeepMind blog.

See the original post:

Google's Artificial Intelligence System - TechMalak (blog)