InspiroBot: The AI That Fails Upward
The origins of InspiroBot are a bit of a mystery, but it's clearly not trying very hard to pass a Turing test. In fact, the bot's viral brilliance is in precisely the awkward juxtapositions its algorithm accidentally creates.
Of course, InspiroBot has gone viral, with nearly 11,000 Twitter followers and widespread news coverage across the July 4th weekend. IFLScience! (where Melissa spotted it and sent it my way) and Nerdist both had description stories up last week, and Daily Mail has done a more in-depth story since.
Two things are interesting to me about InspiroBot:
First, it's an example of how AI doesn't necessarily have to act convincingly human to be successful. InspiroBot succeeds precisely because it fails at being human-like. And because of that, it does things humans wouldn't think to do. Those things are entertaining and memorable. That's an overlooked aspect of AI that could be even more transformative than replacing what humans already do.
Second, sometimes the Internet is a slow burn, and influencers are increasingly important to success. Despite what the Daily Mail article says, InspiroBot is not new. It's actually been around since 2015. In fact, TechCrunch did a pretty straight-up story on it then, but I don't remember that seeing this kind of attention. The bot is hot this holiday weekend simply because the right influeners picked it up, and others followed suit. IFLScience!, Nerdist and Daily Mail are all most likely running the story because one of them saw another do it first.
The Internet has a short attention span. Not everything that's "new" actually needs to be new.