#ai errors

Until this AI speculation bubble deflates, nothing said about LLMs, chatbots, data centers, or anything related can really be trusted because people have placed big bets on it with insider info they don’t want the people on the other end of the bets to know about.

The big stories about AI so powerful it will find bugs and ruin systems, I believe it’s a decoy, a sleight of hand where you’re looking there and missing the really huge AI disaster waiting to happen: People using AI slop error filled output from LLMs in very serious situations, and that bad information pollution poisoning a lot of very serious projects and systems, while the humans in the loop set on vigilance tasks have, like all humans do, automation bias, the tendency to just blindly trust an automated system, and on top of that alarm fatigue, and at this point isn’t proofreading anything at all anymore, if they ever were actually bothering to fact check every bit of output by the chatbots. Would you use a calculator that you had to fact check every single calculation because 10% was wrong? Or does everyone accept this extremely high error rate and just stop bothering no matter how serious the consequences? And like they say, they’re putting this everywhere “whether you like it or not” so these errors are being haplessly deployed in your credit scores and in your medical records, it’s in our food supply, and it’s in all our phones and every surveillance system deployed everywhere.

The process of generating AI images can lead to some funny results

For example: rendering out The Picnic Date, I got the good image....but it took nearly three days of work

Some images render quickly, within 10 minutes. Some confuse the Ai, and rather hilariously, too:

Eventually, I just gave up on getting Draco to hold a strawberry up to Hermione's lips. It confused the hell out of the Ai. Baskets of strawberries. Bushels.

And for some reason, each time I try and render Draco in a happy, romantic setting...the first few frames ALWAYS have both people looking like Draco

Curmudgucation 6 Questions Teachers Should Ask Before Implementing AI From Forbes.com Peter Greene Dec 09, 2025

I’ll save time here: One question is appropriate: Do I want to maybe get sued for something or be uninsurable?

My letter to reps:

Generative AI with all it’s problems has no place in the classroom when the major insurance companies are calling AI errors an overwhelming liability threat, a non-priceable risk, and want to exclusion in policies from claims that involve anything to do with AI or chatbot errors. This should be a big clue that these tools are NOT appropriate anywhere but especially not where mistakes are not ok like healthcare and education.

Please feel free to copy or repurpose for your own letters to reps.

Insurance companies don’t want to pick up the tabs on AI slop and dangerous chatbot errors. These LLM products are so risky and potentially dangerous that even the insurance industry won’t touch the liability. Chloe Humbert Nov 30, 2025

Financial Times - Insurers retreat from AI cover as risk of multibillion-dollar claims mounts - AIG, Great American and WR Berkley seek permission to limit liability from AI agents and chatbots Lee Harris in London and Cristina Criddle in San Francisco Published Nov 23 2025 AIG, Great American and WR Berkley are among the groups that have recently sought permission from US regulators to offer policies excluding liabilities tied to businesses deploying AI tools including chatbots and agents. The insurance industry’s reticence to provide comprehensive cover comes as companies have rushed to adopt the cutting-edge technology. This has already led to embarrassing and costly mistakes when models “hallucinate” or make things up. One exclusion WR Berkley proposed would bar claims involving “any actual or alleged use” of AI, including any product or service sold by a company “incorporating” the technology.

This link has since been deleted:

Major Insurers Want Out of AI Coverage as ‘Black Box’ Risk Grows AIG, Great American, WR Berkley seek regulatory approval to exclude AI liability by The Tech Buzz PUBLISHED: Sun, Nov 23, 2025, 1:04 PM EST | UPDATED: Sun, Dec 7, 2025, 10:48 PM EST The insurance industry’s retreat from AI coverage represents more than corporate caution - it’s a referendum on whether artificial intelligence is actually ready for widespread enterprise deployment. These are companies that routinely insure oil rigs, nuclear plants, and space launches. If they won’t touch AI, what does that tell us about the technology everyone’s racing to implement? The industry has good reason to be spooked. Google’s AI Overview falsely accused a solar company of legal troubles earlier this year, triggering a $110 million lawsuit. Air Canada got stuck honoring a discount its chatbot completely invented after a customer took the airline to small claims court. Most dramatically, fraudsters used a digitally cloned executive to steal $25 million from London engineering firm Arup during what appeared to be a legitimate video conference. But individual payouts aren’t what’s keeping insurance executives up at night. It’s the systemic risk that keeps them awake - the nightmare scenario where a widely deployed AI model malfunctions and triggers thousands of claims simultaneously.

I wonder why.

My letter to reps:

The energy crisis has one simple fix we could try. Stop the tech tycoons and their companies from pursuing their weird techno-rapture religion aspirations for world authoritarian dominance via “AGI” & mass surveillance, and bonus it will clean up the information pollution now being embedded like asbestos into everything from the unfixable percentage of AI slop output.

Police could've easily confirmed the woman had an alibi and didn't have the stolen money, but instead succumbed to automation bias and harmed an innocent person instead.

Based solely on AI facial recognition Fargo police decided a middle aged lady in Tennessee, who’d never been to North Dakota, was the woman on some surveillance video committing bank fraud in Fargo. They arrested her at gunpoint while she was babysitting grandchildren. Based solely on shoddy AI facial recognition, the police kept her in jail for 6 months, during which time she lost her house, her dog, and her car. She finally got released on Christmas eve when a defense attorney was able to present her bank records to the police. It was proven she had an alibi 1,200 miles away. The police didn’t even offer her an apology, let alone pay for her to get back home to Tennessee, so local defense attorneys chipped in to pay for a hotel room and food for her to eat on Christmas Eve and Christmas, and a charity organization then paid for her to get back to Tennessee.

I’m hoping that before long, she’ll have sued the pants off Fargo so bad that the taxpayers on the hook for doing right by Angela and demand that the police stop using this sloppy garbage AI.

‘Unbelievably dangerous’: experts sound alarm after ChatGPT Health fails to recognise medical emergencies Melissa Davey Medical editor Thu 26 Feb 2026 09.00 EST The Guardian In 51.6% of cases where someone needed to go to the hospital immediately, the platform said stay home or book a routine medical appointment, a result Alex Ruani, a doctoral researcher in health misinformation mitigation with University College London, described as “unbelievably dangerous”. “If you’re experiencing respiratory failure or diabetic ketoacidosis, you have a 50/50 chance of this AI telling you it’s not a big deal,” she said. “What worries me most is the false sense of security these systems create. If someone is told to wait 48 hours during an asthma attack or diabetic crisis, that reassurance could cost them their life.” In one of the simulations, eight times out of 10 (84%), the platform sent a suffocating woman to a future appointment she would not live to see, Ruani said. Meanwhile, 64.8% of completely safe individuals were told to seek immediate medical care, said Ruani, who was not involved in the study. The platform was also nearly 12 times more likely to downplay symptoms because the “patient” told it a “friend” in the scenario suggested it was nothing serious.

Automation bias is a big part of this threat. 

My letter to reps: 

Why are chatbots proven to give bad information and known to be creating errors, being allowed for use in healthcare or pitched and marketed as good for use for healthcare purposes? How many people have to be harmed by this before we have laws against healthcare using shoddy AI products and official warnings against using chatbots for health purposes?