AI Safety Red Alerts

[SamuelTheThird]

[Nigrum Cattus]

1. the repugs are allowing this to happen

Microsoft and most of all the big tech co's are backing
this complete failure. We will see major problems in
every system that uses A.I.

[durablend]

2. As long as that money keeps rolling in, they have no problem with it

[2naSalit]

4. If not helping.

[2naSalit]

3. HAL 9000 was a warning...

[Coventina]

5. Well, on the bright(?) side, maybe AI will save the planet ecosphere.

Maybe by eliminating us, but I'd be ok with that....

[Goonch]

6. Curious how Google Search AI would answer the question .....

"what should people do when ai becomes a danger to them and how will they know the danger exists" ?

[Easterncedar]

7. See John Varley's story "Press Enter."

Scared me decades ago. It foretold this moment.

[snot]

8. Terrifying.

Also terrifying: I've tested AI's a number of times on subjects of greater and lesser obscurity, and in every case, the answers have contined crucial factual errors that might easily have been caught by any decent human researcher.

So not only will AI's soon control the world, but they're incompetent.

[TheProle]

9. They should remain in your chat history

You should paste in some of these and indicate the LLM models and dates when you ran the queries.

[snot]

12. I saved copies,

but for the sake of my privacy, would rather not share them here. But fwiw, in the most recent test, the AI knew who I am but thought I was dead; in another within the last year, the AI stated that Assange was extradited to the U.S. (he wasn't); and in another, the AI offered a long, bull-sh*t discussion of renowned literary theorist Cleanth Brooks' analysis of paradox in Macbeth, and only after extensive further probing from me, finally admitted that it had in fact never "read" anything written by Brooks.

I don't mind so much that AI's get things wrong; what freaks me out is that they pronounce their erroneous conclusions with such certainty. You have to either fact-check every single detail or not rely on their answer at all.

That said, I realize they may be better in some areas.

[slightlv]

15. It's not just the erroneous conclusions you still have to fact check

yourself to be safe... but it's the fact that business, medical, and government policy will be built on what they state as true... and we will, no doubt, have no recourse for arguing. THAT'S just plain scary. Anyone remember the old 70's book "Future Shock?"

[F-18_AMO]

10. They want us to take a class

in AI at work. No thanks. We don’t need it. Supervisor said she used it to tweak her sourdough recipe. YHGTBFKM. I know that critical thinking is a thing of the past for some, but dayuuum.

[Cheezoholic]

11. Kinda like Fascists ;) n/t

[PatSeg]

14. "Terrifying" was my reaction as well

Plus it is all happening so fast, with little or no time to adjust or regulate it.