A group of programming teams have developed bots which have been classified to show a “humanness” factor of 50%.
Put to the test at the 2K BotPrize event, two teams’ bots, the UT^2 bot and MirrorBot, split a top prize of $7000 for being the most human-like.
In the competition, computer-controlled bots created by programming teams from all over the world face off alongside human players in Unreal Tournament. The human players act as judges and tag other players (and possibly bots) with being human if their behaviour suggests human control. After several rounds of combat, the bot that has received the most human tags wins the contest.
Other human players picked up an average “humanness” rating of 40 percent, while the UT^2 bot and MirrorBot achieved a 52% rating – the highest ever achieved by bots.
“A great deal of the challenge is in defining what ‘human-like’ is, and then setting constraints upon the neural networks so that they evolve toward that behavior,” said University of Texas doctoral student Jacob Schrum, who worked on the UT^2 bot.
“If we just set the goal as eliminating one’s enemies, a bot will evolve toward having perfect aim, which is not very human-like. So we impose constraints on the bot’s aim, such that rapid movements and long distances decrease accuracy. By evolving for good performance under such behavioural constraints, the bot’s skill is optimised within human limitations, resulting in behaviour that is good but still human-like.”
If you want to test your skills against the University of Texas’ UT^2 bot, check out the research group’s page.
We might as well roll out the red carpet for SkyNet.