Video game bad guys are about to get way smarter — here’s how
Even the toughest first person shooters give Player 1 an advantage over his CPU opponents; the ability to somewhat predict the actions of the computer powered characters and the opportunity to alter tactics when they get stuck.
However, EA is looking to change that by training “self-learning AI-agents” to combat human players in the multiplayer mode for 2016’s smash hit FPS Battlefield 1.
The company’s Search for Extraordinary Experiences Division (SEED) has teamed up with DICE to train the AI agents to alter their behaviour depending on the game scenario.
Related: Best multiplayer games 2018
For example, the agent reacts if his ammo or health is low and goes in search of supplies.
All of the players you see in the video below are part of a single neural network and been trained to play the game from scratch, using trial and error techniques.
In a blog post on Thursday, SEED’s Technical director Magnus Nordin says the tech is still a long way from prime time, but recent tests have shown the self-learning agents holding their own against human players.
He expects AI agents to become part of future games and will provide players with tougher opponents and enemies that adapt and evolve as they learn from experiences of competing against human players.
Nordin says: “I have no doubt in my mind that neural nets will start to gradually make their way into games in the years to come. Self-learning agents aren’t just a good replacement for old-fashioned bots, you can also apply machine learning to a number of fields, such as procedurally generated content, animation, voice generation, speech recognition and more.”
He also says the short term goal is to help the studios collect crash reports and find more bugs.
As you can see from the portion of the video showing the agents gathering to run in circles mid-battle, there’s still a long way to go.
Would you prefer to go against quick learning AI opponents in first person shooters? Or do you enjoy continually outsmarting the computer powered bots? Let us know @TrustedReviews on Twitter.