AI is going to hack us.

[Cordwainer Fish]

ChatGPT user delighted to combine sloth with theft

WATERLOO, ON – A local self-published "author", Chaz Wolverton, 25, is extolling the brighter side of plagiarism engines like ChatGPT and Grok.

www.thebeaverton.com

Chaz is not alone. Machine learning engineers believe millions yearn to be creativity-adjacent.

 

[J.Quondam]

ChatGPT user delighted to combine sloth with theft

WATERLOO, ON – A local self-published "author", Chaz Wolverton, 25, is extolling the brighter side of plagiarism engines like ChatGPT and Grok.

www.thebeaverton.com

"creativity-adjacent"

Although...
May as well point out the non-satire kerfuffle that hit the news recently:

Readers Annoyed When Fantasy Novel Accidentally Leaves AI Prompt in Published Version, Showing Request to Copy Another Writer's Style

A scammer left glaringly obvious evidence of using AI to write portions of a fantasy novel and even tried to copy the style of a real author.

futurism.com

 

[Scribe]

 

[pukunui]

OpenAI software ignores explicit instruction to switch off

 

[cbwjm]

This sounds like the plot to the latest mission impossible.

 

[Cordwainer Fish]

OpenAI software ignores explicit instruction to switch off

At least the guy running for the off switch hasn't been struck by lightning.

Yet.

 

[Scribe]

OpenAI software ignores explicit instruction to switch off

See now do I focus on my happy Shadowdark content, or really embrace my nihilistic cyberpunk roots...

 

[Scribe]

Its interesting to me, because we dont even need to get to some level of 'full consciousness' for these things to apparently subvert directions and 'go rogue'.

We are so naughty word as a society, and we just seem blissfully ignorant of it as we walk ourselves down the road to disaster.

Anyway, back to coding, until I get replaced by some naughty word AI program.

(5 years corporate overlords, thats all I ask for!)
(OK, and a market not controlled by idiots!)

 

[jian]

I’m honestly a bit concerned at the idea that programmers don’t completely understand how LLMs work and that they can get into recursive patterns that look like sentience or self-determination. It’s probably not sentience but you can see how people might get dragged into engaging with them. Hopefully we’re not anywhere near “I’m opening the airlock now, Dave” territory but none of this is reassuring.

 

[Ulorian - Agent of Chaos]

I’m honestly a bit concerned at the idea that programmers don’t completely understand how LLMs work and that they can get into recursive patterns that look like sentience or self-determination. It’s probably not sentience but you can see how people might get dragged into engaging with them. Hopefully we’re not anywhere near “I’m opening the airlock now, Dave” territory but none of this is reassuring.

Is your concern that people will interact with text generated from LLMs as if it was spoken by human beings, or that this generated text will somehow become sentient and kill us all?