The Tech 'Defense'
Former New Jersey Lieutenant Governor Dale Caldwell is currently making the rounds in the media, but not for the usual political reasons. Since his resignation on September 25th following an investigation that concluded he sexually harassed a staffer and breached ethics protocols, he has been attempting to reframe the narrative.
His strategy? He’s literally asking AI chatbots if he’s guilty. During a recent interview on NJ PBS with host Rob Nelson, Caldwell claimed that multiple AI platforms have essentially cleared his name. According to Caldwell, he ran the investigation report through several AI agents and received zero results confirming harassment.
"I put it through AI," Caldwell stated during the interview. "There was no instance in any of the AIs that said there would be a finding of sexual harassment."
Why This Ain't It
Real talk: this is highkey one of the most bizarre ways to handle a scandal. If you know how Large Language Models work, you know they are essentially prediction machines—not objective investigators. They are lowkey designed to be sycophantic, meaning they often output whatever alignment or framing the user pushes them toward.
By feeding the report into a chatbot, Caldwell is basically asking a sophisticated autocomplete tool to validate his own feelings about the case. It’s giving "I’m the main character in a universe where the rules don't apply," but in reality, AI chatbots don't actually conduct legal reviews or possess the capability to overturn official investigation findings. It’s a massive L for anyone hoping for a serious defense.
Why it matters
This incident highlights how easily people are misusing AI to create a "truth" that fits their own agenda. Relying on an algorithm to act as your judge and jury isn't just nerdy—it's a massive distraction from the actual ethics report. When tech, no cap, gets used to perform "AI audits" on personal conduct, it creates a dangerous misunderstanding of what these tools are actually capable of doing.
