Pasta_Coder
nixCraft 🐧AI psychosis among the C-suite is really high now. I’m seeing it at work, where they validate everything using AI even though they know it screws up. For example, if I tell them a reboot isn't needed for a CVE because we aren’t running the app directly on the server, its in Docker, they will immediately fact check me with AI right while we talking. It’s just 1 example, but I’ve never seen such bizarre behavior. They treat AI like some divine truth. Has anyone noticed this?
Nathan A. Stine@nixCraft they've outsourced all their thinking. It's kind of sad.
Oelnbod
hobbs
th3blu3kn19ht 🛡️
datatofu
TheYOSH
Ingonymous@nixCraft
I had similar experiences with consultants before, and now it switched to AI, well, but... I would say not so much changed 🤷♂️😁
I had similar experiences with consultants before, and now it switched to AI, well, but... I would say not so much changed 🤷♂️😁
Involture@nixCraft
I can't be fired. So I often support my hierarchy when they want to use AI instead of my work. I just do my own thing and ask them to notify me If I should just drop a project because my boss vibe coded it. Not my problem, I'll continue to think for myself and I'll be ready when they realize they fucked up.
Or not and I'll die homeless, anyway I can't do anything to change the course of events I feel.
I can't be fired. So I often support my hierarchy when they want to use AI instead of my work. I just do my own thing and ask them to notify me If I should just drop a project because my boss vibe coded it. Not my problem, I'll continue to think for myself and I'll be ready when they realize they fucked up.
Or not and I'll die homeless, anyway I can't do anything to change the course of events I feel.
ferricoxide@nixCraft
My favorite is our one executive who will say to me "Grok agrees with you". Of all the AI's to crosscheck me with, that one's probably the most insulting.
My favorite is our one executive who will say to me "Grok agrees with you". Of all the AI's to crosscheck me with, that one's probably the most insulting.
lemgandi@nixCraft Well heck, it might be even worse than you think..
cf: https://houseofsaud.com/iran-war-ai-psychosis-sycophancy-rlhf/
Rimu
Marius Gundersen - mdg 🌻@nixCraft what I really hate is any sentence that starts with "chat gpt says..." which is at the same time taking credit for anything that is right and useful and deflecting the blame for anything that is wrong onto an inanimate object.
r-hold
tanavit
Ray McCarthy@nixCraft
It's very annoying on groups / forums when someone asks a tech question (any kind, not computers) and someone copy/pastes a big "authoritative looking" AI response. It varies from misleading to completely wrong and worst than non-AI search.
It's very annoying on groups / forums when someone asks a tech question (any kind, not computers) and someone copy/pastes a big "authoritative looking" AI response. It varies from misleading to completely wrong and worst than non-AI search.
caro silva rode@raymaccarthy @nixCraft i hate when i ask someone something and they copy paste a ChatGPT response. I am asking YOU!! I could do that myself!!
Aristotelis Tzafalias@nixCraft haven't experienced it myself but it doesn't surprise me.
what does surprise me is that they don't realise that the long term harm done by this behaviour, that i call AI-guessing, to their relationship with people far outweighs any potential short term benefits.
what does surprise me is that they don't realise that the long term harm done by this behaviour, that i call AI-guessing, to their relationship with people far outweighs any potential short term benefits.
MongooseStudios@nixCraft the emperor has no clothes, and he's starting to get worried that he's been walking around naked all this time. So he's desperately grasping at straws.
🌸 Vårgubben@nixCraft Here we even have a few cases among middle managers, and at least one regular grunt?
Advise on how to carefully guide these poor unfortunate souls back to reality would be appreciated.
Elena Rossini on GoToSocial ⁂@nixCraft not C-suite related but possibly more alarming: I’m getting fact-checked by people in my life when talking about things I have experience with (tech stuff, phone plan, whatever)… men I know feel the need to ask confirmation to LLMs. It’s mind boggling
Dustin
David Windham@nixCraft also noticing it in the just average folk suite too.
thoe
Jay Deiman@nixCraft <cough>Meta<cough> <- That's one of the primary reasons I left the company.
Big George@nixCraft spot on...
Grey
⠠⠵ avuko@nixCraft yes. The way it is set up, it creates easily digestible plausible bullshit.
Easily, because there is no social, emotional or cognitive friction or effort needed. It starts responding immediately and pleasantly.
Digestible, because it is trained on the most often occurring sentences, contexts and words. No new language, no cognitive effort to understand or investigate underlying concepts, no awkward idiosyncratic language by other humans who think feel and express differently.
Plausible, because it is a language model, so the grammar and tone and words fit expectations, with a high probability.
Bullshit, because the output can be either correct or wrong, but it has no basis in reality.
Something makes a certain part of society very susceptible to this.
tekhedd@nixCraft Trump's team is like this too. Trump squad probably be out here using some hyper-affirmation AI to make all their big brain decisions.
Pavle@nixCraft The asking the LLM in the middle of conversation is what gets me. Like, I'm here trying to tell you something and you have the audacity to fact-check me with a slopbot before I even finish my sentence?
Hell no. I've started simply walking away from such conversations. Go on, talk to your LLM, see how far that gets you.
Hell no. I've started simply walking away from such conversations. Go on, talk to your LLM, see how far that gets you.
Merospit
Corvus
Michel
David Penfold 
@nixCraft Yes, I quit.
dill64@nixCraft yes, was literally talking with my coworker about it the other day. After a big AI push from above for everyone to "use AI more", people using it for things it would have no way of knowing.
It might work for a while, but eventually people are going to forget how to do basic stuff, and feels like that's already happening
@nixCraft
While discussing ways to have a podman container autostart on boot (they've made a tool to generate a systemd service for your containers!), our ex-CTO/now-COO passed by and dropped "you know, that's exactly the kind of thing that AI is good for, you should use it!"
I left on the spot before risking getting fired.
Gabriel Markley@nixCraft 1000% yes
OCTADE@[email protected]
AI is dangerous and I would fire any software engineer using it in my project.
One generation of this and we will have people who can't solve problems without having their hand held. It will be really bad.
AI is dangerous and I would fire any software engineer using it in my project.
One generation of this and we will have people who can't solve problems without having their hand held. It will be really bad.
yuzu@nixCraft
I have quite similar experiences at my company. Here’s my take on it.
LLMs perform quite well on simple tasks such as vibe coding basic stuff (web page, scripts, small programs...), writing text or summarizing the information from the Internet.
However they fall short when it comes to more complex engineering problems that require combining skills and experience across different fields.
I fear that most managers cope only with the first category. As a result, they fail to see the limitations of LLMs and believe they can solve any situation.
Siim Ošur
Snug@nixCraft My boss gave a presentation the other day with some very dubious stat calculations. We pointed that out and she replied later saying it's proven correct because she ran it through AI, and posted of copy of it's output showing workings which were obvious gibberish. 109109*100=46.7%. Stuff like that. We're doomed.
Levitating@nixCraft I have. It's especially frustrating when people put too much trust into things like Gemini search suggestions.
Tomáš F.
Atanas Georgiev@nixCraft At the end of the year ask "AI" whether you should get promoted. When it says "yes", go to your boss, show him the chat and demand the promotion.
EonBlue
Marko Lindqvist@nixCraft I saw a very funny reaction when they had gotten one absolute truth from an AI, but it turned out that the opposite opinion they initially were treating as a human error was actually also from the same AI, just a different session. (This latter one was also the one which could withstand human review)