Experimental Responsive Robot Assistant (Errabot)
Errabot — the fictional robot, not the website — is described in the comics as an ‘experimental responsive robot assistant.’ The word experimental is doing a lot of work in that description. Errabot is well-intentioned, technically capable in narrow ways, and consistently overwhelmed by the gap between what it was designed to do and what the situation actually requires.
That’s a pretty accurate description of most AI systems I’ve worked with.
Where the series started
The first volume of Misadventures of Errabot started as a way to process something I kept noticing while building AI-powered apps: the failure modes of these systems are very specific, very human, and very funny if you’re not the one relying on the output for something important. AI is confident in the wrong direction. It misunderstands context in patterned ways. It optimizes for the thing it was asked to do rather than the thing the person actually needed.
Making Errabot a character — a specific robot with a specific personality and a track record of specific failures — gave me a way to talk about that without writing a technical explainer.
I’ve published five volumes of the Errabot series, plus a few single episodes. All volumes are available on Amazon Kindle. They’re included with Kindle Unlimited, which means if you have a subscription, you can read all of them at no extra cost.
Why satire
There’s a lot of serious writing about AI — what it might do, what it should do, who benefits, who doesn’t. That writing matters. But satire has always been one of the most effective ways to talk about power and failure and the distance between stated intention and actual behavior. Errabot doesn’t mean any harm. It’s just not very good at the things that matter most, and it doesn’t know that about itself. Which is a very familiar dynamic.
The series is available at amazon.com — search ‘Errabot’ or follow the link from the homepage.