this post was submitted on 12 Feb 2026
129 points (100.0% liked)

Fuck AI

5728 readers
1029 users here now

"We did it, Patrick! We made a technological breakthrough!"

A place for all those who loathe AI to discuss things, post articles, and ridicule the AI hype. Proud supporter of working people. And proud booer of SXSW 2024.

AI, in this case, refers to LLMs, GPT technology, and anything listed as "AI" meant to increase market valuations.

founded 2 years ago
MODERATORS
all 26 comments
sorted by: hot top controversial new old
[โ€“] Denjin@feddit.uk 14 points 6 hours ago

Subject: Fire. Dear Sir/Madam, I am writing to inform you of a fire that has broken out on the premises of 123 Cavendon Road... no, that's too formal. Fire - exclamation mark - fire - exclamation mark - help me - exclamation mark. 123 Cavendon Road. Looking forward to hearing from you. Yours truly, Maurice Moss.

[โ€“] subignition@fedia.io 49 points 11 hours ago (2 children)

When the alarm goes off in your building... you evacuate. Doesn't matter if it's a test or not.

[โ€“] NOPper@lemmy.dbzer0.com 1 points 1 hour ago

Read the post, the follow up explains she was already outside at the designated meeting point per policy and just making a Slack thread to talk about it.

[โ€“] surewhynotlem@lemmy.world 27 points 11 hours ago (1 children)

Exactly. It's literally the point of a test.

[โ€“] Mouselemming@sh.itjust.works 9 points 7 hours ago (1 children)

It's literally the point of a fire drill.

A scheduled test of the equipment, with the alarms going off randomly all day as the tester gets to them, doesn't require you to evacuate.

Such a test is always supposed to be preceded by ample warning notifications, so people will be extra careful not to set real fires while the building is vulnerable.

[โ€“] Hawke@lemmy.world 7 points 7 hours ago* (last edited 7 hours ago)

And also is supposed to be monitored with people actually watching for fires so they can manually evacuate people if a fire does break out during the test. (Requirements may vary by jurisdiction of course)

[โ€“] tynansdtm@lemmy.ml 40 points 12 hours ago (1 children)

AI offering unsolicited advice on emergency situations is a fresh, new level of hell.

[โ€“] atrielienz@lemmy.world 14 points 10 hours ago (1 children)

I didn't realize until I read your comment that the AI was integrated into Slack and told this person that they didn't need to evacuate without them specifically asking the AI for advice.

On the other hand, this does show that anything typed into that slack channel is treated like a query. Which is also terrifyingly stupid.

[โ€“] tynansdtm@lemmy.ml 2 points 2 hours ago (1 children)

One also has to wonder, did they message on their phone from a safe location? Or did they hear alarm bells and their first instinct was to get on their computer and start typing?

[โ€“] atrielienz@lemmy.world 1 points 2 hours ago* (last edited 11 minutes ago)

My hope is they did this after they evacuated. But honestly could go either way.

[โ€“] theacharnian@lemmy.ca 42 points 12 hours ago (1 children)

I came here to say that only an idiot would ask an AI to explain a fire alarm, but then I actually read the post and holy shit. It was an AI agent responding on slack. So yea, this is serious.

[โ€“] ivanvector@piefed.ca 9 points 10 hours ago (2 children)

I mean, they're still an idiot. When the fire alarm goes off you get the hell out of the building, not start a group chat on Slack.

[โ€“] Zorcron@piefed.zip 1 points 51 minutes ago

OOP clarified that they did leave the building first.

[โ€“] theacharnian@lemmy.ca 10 points 10 hours ago

That's assuming too much, though. One can post on Slack from their phone outside of the building. This is all we know:

A colleague wrote to a Slack channel 'Fire alarm in the office building', to start a thread if somebody knows any details.

Regardless, moon, not finger.

[โ€“] psx_crab@lemmy.zip 13 points 10 hours ago (1 children)

Someday, AI will kill us.

It won't. The lack of common sense will, and in this case, starting a group chat about the fire while the alarm is ringing.

[โ€“] frosty@pawb.social 7 points 6 hours ago* (last edited 6 hours ago)

According to the original poster, they clarified the slack chat was started after the person had evacuated the building, and the AI was configured to glean prompts from the chats and respond automatically if it deemed it could help.

https://mastodon.online/@tagir_valeev/116059238703673745

[โ€“] Triumph@fedia.io 15 points 11 hours ago

It's going to give you the "most likely" answer, stated with utter confidence. Alarm tests happen more often than real alarms.

But then ... shouldn't it be saying "whether or not this is a test, you should follow your building's emergency evacuation procedures immediately"?

[โ€“] pkjqpg1h@lemmy.zip 3 points 11 hours ago (1 children)

Never use AI for advice...

[โ€“] samus12345@sh.itjust.works 6 points 7 hours ago (1 children)

They didn't. It was unsolicited.

[โ€“] pkjqpg1h@lemmy.zip 4 points 6 hours ago

hmm, let's correct: never trust AI for advice

[โ€“] 13igTyme@piefed.social 4 points 12 hours ago (2 children)

We should just treat AI like I treat Jim Cramer. Do exactly the opposite of what it says.

[โ€“] pkjqpg1h@lemmy.zip 3 points 11 hours ago (2 children)
[โ€“] Nollij@sopuli.xyz 2 points 5 hours ago (1 children)

He's a guy with a show about stock/investing advice, widely reported as an expert in his field, with a less than 50% success rate.

OP is literally, technically correct that doing the opposite is a better strategy than actually following his advice.

[โ€“] pkjqpg1h@lemmy.zip 1 points 5 hours ago

doing the opposite is a better strategy than actually following his advice

๐Ÿ˜‚

[โ€“] lost_faith@lemmy.ca 4 points 10 hours ago

The absolute best stock adviser on tv, so long as you follow 13igtyme's advice lol

p.s. NOT FINANCIAL ADVICE