(Disclaimer: this is a work of fiction. Learn to take a joke; you’ll live longer.)
A CRZ-3001 unit was replaced on Friday after it started malfunctioning. The CRZ-3001, better known by its market name Cynthia Ritchie, was replaced by the Corporation technicians immediately after an alert was issued by its Monitoring Officers.
“Yeah, the unit (serial #212) was acting up, so we had to change it,” said Khalid Hassan, chief technician, Zone 7(b). “We tried to fix it but it was taking too much time, so we just replaced it with a new one. I hope 213 last a bit longer or else the folks at AI-HQ would have to do away with the CRZ-3001 completely.”
“Look, it’s task is simple. Its neural nets have been trained on a large enough data set of twitter conversations to pass of as a completely agenda-less real human and it does a pretty good job at that. I mean look at the detail, that skin looks so real, as does that friendly banter on twitter,” he said. “Not enough to quite overcome what we in the robotics community call the Uncanny Valley, but still does a pretty decent job.”
“The problem with 212 and its predecessors was that when engaged by many on Twitter, it used to start revealing its source code,” he said. “When someone was speaking on the horrors of the ‘71 war, it would blurt out how you shouldn’t judge because you weren’t there, etc etc.”
“Sometimes the folks at AI-HQ would take over the twitter handle itself,” he said. “but that was also a problem because their English was horrible, revealing themselves immediately.”
“The final blow came when it tweeted ‘I wonder where so-called “liberal” writers/activists, who are complaining of “white privilege”, “colonization” – from where do they get their paychecks? Where do they live? Ideas?’”
“Not revealing the source code was its basic job,” I said. “I hope heads roll at AI-HQ for these glitches.”