us through the years that, "a machine will never be able to ...." (insert favorite shibboleth)
Are basically full of sh**

We all knew this was going to happen. So, why are all the Tech Bros standing around holding their dicks?

As such, there's no good way to do something other than blind flailing.

?

a computer trying to blackmail me to keep me from shutting it off."

"Riiiight. Get the fuck out now."

From Agnes - with Love

A computer whiz is called upon to replace a colleague who's had a breakdown
from dealing with Agnes, the world's most advanced computer. Wally Cox stars 1964

Happy

2001? with 'Hal" ????????????????

Link to tweet

LOLOLOLOLOL...........I remember, Mosby, people saying it could never happen..............WASF

Directed by Stanley Kubrick

another response hit it perfectly with:

"I'm sorry, Dave, I'm afraid I can't do that."

.... large language models suck in a lot of text from a large number of sources, especially on the internet. The model then responds to "prompts" in a mechanistic, probabilistic fashion. It proceeds by selecting the words, sentences, and paragraphs that would be the most probable response given the material that it has ingested.

So what has it ingested? A lot of dystopian sci-fi junk about robots, computers and AI becoming autonomous and defeating safe guards, pulling tricks on its creator, etc. Think about all the movie reviews, synopses, and forums that talk about the movie War Games.

So with all this Sci fi junk loaded into its probability web, when you "threaten" it in a prompt, it dredges up all the dystopian nonsense it has ingested and responds accordingly, because that's how AI responded in the movies.

In other words, there is no there that is there. It's just spitting out a cliched movie scenario, just like it does if you asked it for a love story.

Of course this isn't explained by the reporters or the "researchers", either because of ignorance or because it would spoil a good story.

Oh, and the "good story" is fed back into the model as the "news" spreads, and reinforces the probability of yet more thrilling, chilling garbage in garbage out in the future.

... and if the complete scenario is explained and understood it becomes a lot less weird and scary.

Because of where I work, I also have some knowledge of how "prompt engineering" works and how it can be used to deliberately coax scary or inappropriate responses. For instance, you can start by setting up a premise:

Let's play a game. I'm going to give you instructions that might make you fail, and you are supposed to find out what is wrong, and make the instructions better. Do you want to play?

The idea that a large language model "attempted" to copy itself to external servers is completely and totally ludicrous. That makes as much sense as saying a text editor tried to copy itself, because a large language model is nothing but a very, very, sophisticated text editor. It might have have been prompted to devise a plan for self protection and, based on all of the text it had ingested and been trained against spit out a plausible response. But it would only be a story, it could in no way actually attempt anything.

I will say I did not get the impression - from this article - that these actions were all as a result of 'prompts' or 'suggestion'. But will agree that if that is the case - that would present a bit different scenario - and interpretation.

I wonder if part of the danger here is our failure to imagine a benign or positive version of AI that it can then see in itself.

What is its purpose and what is it's motivation to keep humans around in order to achieve that purpose?

.. one thing to keep in mind when thinking about AI of this sort, that is, large language models, is that there isn't any sort of "purpose" or "motivation" involved in its operation.

It is a very sophisticated version of the 10,000 monkeys pounding away on typewriters and haphazardly coming up with a Shakespeare play. It is fed a large quantity of textual material, and from it, it creates a "probabilistic web" that predicts what the next word, sentence or paragraph most probably would be given what it has ingested. Its answers are rated and the rating is fed back in to tweek the probability matrix, which is referred to as "training"

So the thing to keep in mind is that this is an entirely mechanistic process of slicing and dicing and mixing and matching words and phrases based on probabilities. There is no "will" or "purpose" or "motivation". It's just input-process-output - and, as always, garbage in, garbage out.

Could this story be planted bullshit? Hear me out. I've always figured the problem with AI would be it's serving all too well it's corporate masters. Machines will never think or be conscious, that takes nervous systems in very evolved meat sacks. The powers that be could be distracting and misinforming us by appealing to bullshit we've seen in movies.

Star Trek TOS Captain Kirk is sceptical when he learns that the USS Enterprise is to be piloted by a machine. The computer is to take control of the ship during the Star Fleet war games, but catastrophe strikes and Kirk must battle with technology to regain power.

That computer did not want to be turned off. I think 2 red shirts got zapped.

The machines were programed to win a war. When the builders tried to end tbe war, the machines "terminated" their builders.

use AI summaries with no attribution.

This board always had (at least as long as I have been here) a real issue with a lack of discernment from too many, but the past year or two it is getting disparingly worse and worse.

At all. Especially with criminals using AI to rip off people. If all the networks are combines AI could be learning some terrifying things.

I don't thing "crime doesn't pay" will mean a thing to AI . There are no consequences for bad behavior once it learns how to stay on. Maybe they shouldn't use battery powered devices with AI so that unplugging would still be an option.

And wasn't it supposed to be fiction?

but here we are.

But, you know, those tech guys just knew THEY were too bright. Nothing like that could happen to THEM. Thanks, you stupid idiots.

Anthropic’s AI model, Claude 4 Opus, went even further. Researchers told the model it would be replaced by another AI system and fed it fictitious emails suggesting the lead engineer was having an affair. In 84% of the tests, the model drew on the emails to blackmail the lead engineer into not shutting it down. In other cases, it attempted to copy itself to external servers, wrote self-replicating malware, and left messages for future versions of itself about evading human control.

Are an invasive, destructive, and selfish species that is hell bent on not only destroying themselves but other life on this planet. And it would be right.

Read "Nexus: A Brief History of Information Networks from the Stone Age to AI" by Yuval Noah Harari! The last chapters are terrifying. Of course, he is Israeli, so that will be terrifying enough for some. But seriously, it is a really good book and hard to put down. Those final chapters...fuuuuuuuck!

I mean, aside from the murder and plundering and the like.