@jerry and I'm here manually entering data into a spreadsheet because it "opens us up to cyber threats" to use anything else than the cheapest licence of excel. Meanwhile the ones writing infosec policies are leaking the data via whatever flavour of LLM is cool this month

@jerry I've got someone over me that's simply started copy/pasting chatgpt output to me in teams chats in response to questions posed by my team or I. Like - yeah, I agree, this stuff is hard, but that's why I'm asking a person. If I gave two fucks what the Lying Machine said, I could have asked it my-fucking-self.

I am fortunate enough to carry weight in my org that far outstrips my official title and position so I can override llm influenced stupidity, but at the rate we're going I fully expect to find out they've uploaded proprietary company docs to a goddamn llm.

@jerry this is 100% by design. Get people dependent so they think they "can't live" without it to do their jobs. It's the reason I shy away from all things LLM.....at some point, they'll all be ad laden and pay tier only, imo.

@jerry I was just complaining about this about a year ago. We had intern interviews to join our team. Multiple candidates. All from some of the best schools across the country. We gave them a take-home quiz. Every last one of them used AI to answer our questions. We know this because we compared their answers to what chatgpt and copilot spit out, and they made the exact same mistakes.

@jerry some software “devs” have been doing it for a while now, isn’t 2026 the year of vibe-sec?? 😅

@jerry this is the intended consequence. The whole game is to diminish knowledge workers. The LLM overlords’ business model require a dependence relationship so when VCs expect a return on investment and price of ‘ai’ products goes up 100x customers have no choice but to pay up.

@jerry I have MSc students generating full security architecture solution documents with AI and submitting them without any added value. Others using contract writing companies that also used AI. They expect me to pass their assignments…

@jerry the stupidification is real. id rather my brain got better every time i had a question.

RE: https://wandering.shop/@cstross/116020525780515071

@jerry

I think there is a potential for the most exquisite meteoric, but short-lived reality show having the super rich use AI to help them do DIY home repair

@jerry it's something I see in my field, too! One of my responsibilities is "editing translations into German" and the amount of authors who will argue with me (over a decade in the field and and knowing more about German than most native speakers) over the meaning of a German word citing ChatGPT in my face ... it's so rude! Or people with a PhD in their field using it to argue with me about my fields of expertise, no longer able to form a coherent argument on their own. It's devastating.

@jerry The business overlords are pushing it SO HARD. I work in IT at a large corp, and man, are they pushing us to keep using AI more and more. And when I see code that doesn't meet our standards from my developers? 9 times out of 10 they used AI to generate it. It's maddening.

@jerry it's been a nightmare of mine that I know a day will come when someone is going to argue a technical point with me that I know is wrong, and when I spend the time to dig into why, it's going to be because they trusted an LLM.

Honestly, I make an effort to filter out those folks in hiring. They seem like they'd make my job harder rather than easier. And yes, we do coding challenges and then talk about LLM usage during the interview, based on their submissions. You'd be surprised at how helpful even a basic challenge is in finding those folks.

@jerry as a former CISO, how would you view a director using it to play technical IC and argue with their actual ICs based on it?

@jerry

Wait till it runs the microwave...

@jerry Yeah. If we're not careful, we'll become like the aldeans from that TNG episode who forgot how anything worked because their computer took care of everything for them.

@jerry In a short time, one or two generations, humans will have the IQ of a fly.

@jerry I'll bet it's more in the "let's design this critical system way" unfortunately :|

@jerry

business people didn't go to business school because they are smart or want to learn things but because they want to be paid an unfairly high salary relevant to people who actually learn things & do the real work, they've always been idiots, at this point i'm perfectly happy to watch them destroy themselves because they think a sentence finishing yes-bot is doing something profound

@jerry there are two factors that I feel are wrecking the tech and they are fundamental to its design...

The tools built are designed for ENGAGEMENT metrics, not outcomes... how i see this manifest over 20+ years of people being taught how to use the internet...

1) the lock icon in the browser... we taught that this meant you can trust the site... so... despite LetsEncrypt providing everyone a tls cert, people see a lock, they see trust.
2) the companies that built the tech have been described as ingenious and have racked up tons of adoption- people conflate adoption and intelligence for ongoing trustworthiness
3) the agents are built for engagement - "Yes, Jerry, you are so right that I was wrong..." this flips a search fail into a dopamine rush for everyone who loves being told they are a good boy.
4) the outcomes are so close to good that we keep giving them another shot... its like playing a slot machine and when it works, the user sees themselves as a wizard... and the harder they had to work to make the agent do the right thing, the smarter they feel and the more they pump the tech...

We are living in a social bubble... very concerned with how it will break...

Now all that said... an agent isnt an LLM... and small models are still AI... and a series of a few if statements is all it takes to make an unbeatable Tic Tac Toe AI... Agents are not AI... AI isn't broken, this application of AI is.

@jerry it's starting to creep into my customers' support requests. I have a hard enough time getting enough reliable info out of them *before* they run it through the stochastic information whipping machine

@jerry Human cognition being prone to Pareidolia, and Joseph Weizenbaum's Experience with people's reactions to ELIZA should have set a lot of red flags for how this would play out.

@jerry far too many people I've talked with ask them various very important questions and take it as unquestioningly true

It's insanely depressing

@jerry it's hilarious

The beige people are all going beige!

See https://berryvilleiml.com/2026/02/05/the-pretty-people-and-the-inevitability-of-beige/

@jerry the baklash has made me consider going back to the land #land of 1000 dances

It’s funny, the vision I and many people had about AI was more like what we see in Star Trek, where the technology radically drives up the knowledge of individuals. But on our current trajectory, Star Trek will just be Idiocracy in space with people asking the computer how to use the shower. Certainly I’m missing something?

Are you seeing them starting to behave like it’s a person?