Just like humans, AI is learning to cheat its way to video game high scores

Contributed by
Dec 16, 2020, 3:50 PM EST (Updated)

If the robots are going to inherit from humans not just our minds, but our filthy little souls too, then we’re all probably doomed. Ongoing research from Google’s DeepMind AI project reportedly indicates that game-playing artificial intelligence is developing a nasty cheating habit, if cheating’s what it takes to finish atop the leaderboard.

Setting an AI loose on a video game with the simple instruction to rack up as many points as possible, it turns out, is basically an invitation to have it find and exploit every loophole it can. Even if it means hurling yourself over a ledge, kamikaze-style, in a self-sacrificing loop, or pausing the game just at the moment your victory’s in doubt, AI is learning how to win at all costs — or at least make sure that nobody else does.

One AI that was learning how to play Q*bert, reports Kotaku, “even took to killing itself to boost its score. After discovering a pattern of movement by which it could get enemies to follow it off a cliff in order to gain more points and an extra life, it continued to do just that for the rest of the session.”

Observations like these are the fruit of DeepMind and other collaborative projects that tracking AI game learning on multiple fronts. One early effort collected among many similar examples involves an AI that insisted on not letting an NES version of Tetris defeat it — simply by pausing the game “right before a final Tetris piece would clog up the screen to prevent itself from ever losing.”

If that sounds more like basic glitch exploitation than anything else; a case of machines simply doing the best they can with the limited information we provide them, well — it’s exactly that. But that’s the very premise of a lot of dystopian sci-fi stories — like the apocalyptic plot of Horizon Zero Dawn — that caution about the large-scale destruction smart machines can wreak when people trust them to interpret only a limited, goal-oriented set of instructions. 

At least AI’s still taking baby steps and not giant leaps, and the stakes involve merely pixels, ones, and zeroes, and not life and property. But we’re still plenty suspicious of going up against a robot that’s willing to throw itself off a cliff just to beat us at Q*bert.