An Intellyx BrainCandy Brief
To err is human. To really eff things up, you need me. That’s just one of the many realistic conditional error statements SorryAI’s DaveGPT™ chatbot can deliver through an ITSM platform or messaging app.
Most AI applications were built to augment human capabilities, and continuously learn and improve their performance over time. But that’s not good enough for Toronto-based SorryAI. Their LLLM (large lame language model), written in dissembler, even when given all the reasons in the world why it could do better, endlessly makes apologies when prompted, giving humans a ready excuse for just about any failure that happens as a result of their newly augmented incompetence and oversight.
DaveGPT’s self-deprecating error-handling routine could come in handy for industry verticals such as banking, to explain why their systems didn’t notice high levels of interest risk; or logistics, when explaining why a train without working sensors or electronic brakes went off the tracks; or in education, so a student who is late delivering an assignment can say “Sorry Dave, the AI ate my homework.”
Developers will also appreciate this model’s versatility for generating buggy code in any language, from inserting simple syntax errors to creating hidden recursive infinity loops guaranteed to max out available system resources and push out that looming release date.
When this analyst asked DaveGPT to explain itself, it quoted a line from famous Canadian Paul Shaffer’s character Artie Fufkin in the classic film Spinal Tap: “Do me a favor. Just kick my ass, okay? Come on. I’m not asking, I’m telling you. Kick my ass.”
Comments
Comments are closed.