The whole experience reminded our senior editor for AI, Will Douglas, of something far less exciting: Pokemon.
Back in 2014, someone set up a Pokemon game in which the main character could be controlled by anyone on the internet through the streaming platform Twitch. Playing was as complicated as it sounds, but it was incredibly popular: at one point, a million people were playing the game at the same time.
“It Was Another Weird Online Social Experiment Picked Up by the Mainstream Media: What Does It Mean for the Future?” Will would say. “Not much, it turns out.”
The frenzy about Moltbuck struck a similar note with Will, and it turned out that the source he spoke to was also thinking about Pokemon. Jason Schlotzer, at the Georgetown PSAROS Center for Financial Markets and Policy, sees the whole thing as a kind of Pokemon battle for AI enthusiasts, in which they create AI agents and deploy them to interact with other agents. In that light, the news that many AI agents were actually being directed by people to say things that made them feel emotional or intelligent.
“It’s basically a spectator sport,” he told Will, “but for language models.” “
Will wrote a great piece about why Moltbuk wasn’t the glimpse of the future it was made out to be. Even if you’re excited about the future of agent AI, he said, there are some key pieces that Moultbook makes clear are still missing. It was a chaotic forum, but a truly helpful hive mind would require more coordination, shared goals and shared memory.
“More than anything else, I think MultBook was having Internet fun,” Will says. “The biggest question that leaves me now is: How far will people push AI just for laughs?”
Read the full story.