This Halloween, the MIT Media Lab is launching BeeMe, a dystopic tale about an evil artificial intelligence named “Zookd”. Starting at 11pm EST on October 31st, the story’s protagonist will be tasked with stopping Zookd in real time. There’s just one twist: in order to face off against Zookd, the protagonist will relinquish all free will, leaving their every action over the course of the story up for popular vote.
According to BeeMe’s project page, the game is designed to “push crowdsourcing and collective intelligence to the extreme to see where it breaks down”. The system itself is relatively straightforward: viewers can log on to the website and submit recommended actions for the protagonist to do, as well as voting alternate proposals up or down. After a certain amount of time has passed, the remote controlled human is charged with carrying out the most popular choice. In a lighthearted example, BeeMe’s Twitter account shared an example of BeeMe volunteer Evan staring intently at a stuffed giraffe as a direct result of the dubious wisdom of the crowd during a focus group leading up to the official release.
Here are some moments from our Friday night session: Stare intensely at the giraffe. pic.twitter.com/XM7maCeXEx
— BeeMe (@beeme_mit) August 6, 2018
Cognizant that not all requests would involve staring at stuffed animals, project lead Niccolo Pescetelli explained to Business Insider that “anything that violates the law or puts the actor, their privacy, or their image in danger is strictly forbidden…anything else is allowed.” The experience’s trailer doubles down on this ominous spin on things, with the “Wisdom Program Communications” program seeking verbal assent from the “Agent” before handing over bodily autonomy to the collective BEEs (Behavioural Enhancement Entitites).
Handing over individual control to collective forces has been a subject of fascination for some time, with Twitch Plays Pokémon seeing if the comments section of a Twitch channel could be capable of beating Pokémon Red one move at a time. While the game’s initial “Anarchy” mode treated all commands sequentially, the game’s eventual introduction of “Democracy” mode subjected all decisions to a public vote. BeeMe‘s structure bears a close resemblance to Twitch Plays Pokémon‘s Democracy mode. While the Twitch incarnation of this idea took collective commands and input them into a Nintendo controller, BeeMe will be feeding those commands into a trained actor.
This past year’s MIT Mystery Hunt also explored different variations of this idea with their Inside Out-themed puzzle hunt. For Twitch Plays Mystery Hunt, puzzle hunters collectively played a single player game in Anarchy mode. Completing that puzzle unlocked Under Control, where one player on each team was sent as tribute to give up their free will and follow instructions sent via YouTube Live comments to navigate an augmented reality world that was only visible to people watching the stream. The Hunt’s grand finale required players to follow a series of intentionally convoluted instructions to remotely control “Terry” as she navigates the halls of MIT while livestreaming her progress via camera rig.
BeeMe takes things a step further, challenging viewers with an open-ended playground to navigate, with almost no limits in place to curb the crowd’s impulses, for better or for worse. The question here isn’t merely one of seeing whether collective intelligence can identify the optimal courses of action, but whether they can impose enough structure to act upon those courses while resisting the temptation to stare at stuffed giraffes, or worse. As the game’s promotional materials repeatedly asks, “How does it feel to be the Internet?”
Pescetelli estimates that the Halloween hijinks should take around two hours to play out, although the full length is ultimately up to the choices made by the project’s viewers. But the real question remains: will viewers be able to Twitch Plays Pokémon their way into saving the world from an evil AI, or will their capricious choices cast viewers as the real villain of the story?
Go to BeeMe.online to join in the experiment at 11pm on October 31st, and follow @Beeme_MIT on Twitter. But try to behave – there’s a lot at stake.