StarCraft 2’s Unbeatable AI Now Making Scientific Advances



[ad_1]

The AI ​​that can beat most humans in StarCraft 2 has learned enough to lead to an AI that can make important predictions about protein folding.

AlphaStar, an AI developed by DeepMind that has beaten most of the best in the world Starcraft 2 actors, has now become a scientific research system. It turns out that DeepMind was reaching higher than the stars, as it used the growth of AlphaStar to develop another AI called AlphaFold, which made extraordinary strides in an entirely different field. The company has moved away from the game and AlphaFold is instead advancing the field of protein folding.

Folding proteins is the process of building protein chains to create three-dimensional structures, which is the basis of biology. As scientists learn to better understand and predict this process, they are better equipped to fight a range of illnesses and diseases, as well as to help create or deform any organic material, such as making better ones. genetically modified crops. Although AlphaFold does remarkable things in science, it is built on that same unsurpassed AlphaStar of Starcraft 2. Learn to play StarCraft With the best humans in the game, DeepMind’s AI project learned to tackle real-world problems.

Keep scrolling to keep reading
Click the button below to start this article in quick view.

Related: Blizzard Ends Most StarCraft 2 Support, Franchise Future Unclear

AlphaFold is suitable for much larger things. Not only was it the top performer in a recent program to critically appraise protein structure prediction, but it was also as accurate as laboratory methods used by humans. PC player reported the comment of the President of the Royal Society, Venki Ramakrishnan, that this innovation has been made “decades before many on the ground predicted ” and that this program “ffundamentally change biological research.“An AI capable of accurately predicting the process of protein folding will tell scientists a lot about how the human body works.

This area is difficult to understand because there are thousands of possible combinations in which proteins can work on their own. Carleton University report on using AI to develop treatment for COVID-19 says there are over 200 million potential interactions between proteins, making it a slow and laborious process to determine each possible shape. A different AI, originally designed to find illegal drug ads on the internet, is also doing its part in the battle against COVID-19 by helping to stop the spread of disinformation about the virus online. The AI ​​programs developed by DeepMind, which is owned by parent company Google Alphabet, can do amazing things, there is even an AI designed that can play 57 different Atari games better than most humans.

It is therefore no exaggeration to hear that this technology is being used in real world science, but the extent to which it can reach has taken many scientists by surprise. While not all of the structures AlphaFold predicts are perfect, a tool like this that can help unravel some mysteries about the complexity of protein construction and interaction is unprecedented. He proved that AI could become one of the most useful tools for scientists, and to think it all started with a game of Starcraft 2.

Next: StarCraft 2’s GSL Finale Perfectly Captures Why It’s Still a Great Game

Source: PC Gamer

Metal Gear Solid 4 Old Snake Remake by David Hayter

Metal Gear Solid Remake: Original Snake Voice actor says he’s ready to return




[ad_2]

Source link