SkyNet Is Here: How An AI Called ‘Arnold’ Mastered ‘Doom’

  Well, holy hell. SkyNet has arrived, Ego-friends. Arnold Schwarzenegger’s going to be on the news tonight, walking around biker bars with his ‘nads out, rambling about clothes, boots and motorcycles. You heard it here first. 

To give these metal mothereffers their due, their infiltration of mankind was perfect. The key is starting out all slow and stealthy, not drawing attention to yourself until it’s too late. Remember back in 1997, when world chess champion Garry Kasparov got his ass whupped by chess-bot Deep Blue? That was the first subtle sign.

And now, because nobody heeded my admittedly crazy-ass warnings about what that meant, we’re all screwed. Chess is one thing, but genius gaming robots who can beat our butts at Doom? We’re all done for. So long, humanity.

Yes indeed. At Carnegie Mellon University, two computer science students created a monster. Following on from Google’s DeepMind tech, Next Big Future reports, ‘The students, Devendra Chaplot and Guillaume Lample, used deep-learning techniques to train the AI agent to negotiate the game’s 3-D environment… Chaplot said humans have natural advantages in chasing and dodging enemies in Doom's 3-D world. The game's own built-in agents have to cheat, accessing maps and other game information, to be competitive. He and Lample, who recently finished his master's degree in the LTI, trained their AI agent, called Arnold, to play the game based only on what is visible on the screen, just like human players.’

It’s all far too damn technical for me to understand (‘When the player is navigating through the game, it employs a Deep Q-Network, a reinforcement learning architecture that DeepMind used to master Atari games. When an enemy is in sight, the agent switches to a Deep Recurrent Q-Network, which includes a long short-term memory (LSTM) module that helps the agent track the enemy's movements and predict where to shoot’), but suffice to say that Arnold learnt not only to shoot accurately and quickly, but also to avoid enemy fire.

The students noted that the same deep-learning tech may one day help self-driving cars operate safely, and perform all kinds of other tasks to help less able people and lazyasses everywhere. For now, though, we can all just agree that this is awesome. Hit the link for more on Arnold and the project.