In 2018, a group of grad students at Georgia Tech published a paper that noted games had often been “an important testbed for artificial intelligence” in the past. AIs have been beating us at chess, checkers, and Unreal Tournament for years, after all. Since that was the case, they argued, it makes sense for future testing of AI to focus on whether they could learn to play tabletop roleplaying games. Games like Dungeons & Dragons would serve as an effective measure of the progress of AI, “due to an infinite action space, multiple (collaborative) players and models of the world, and no explicit reward signal.”

What they didn’t predict was how controversial AI would become by the time that was possible—a topic we’ll come back to later.



Source link