Is chat GPT a danger to humanity?

[Count Steer]

Prompted by much of the speculative fiction ^^^ regarding AI I decided to knock up my next literary submission to Interzone...

The (Slightly Wonky) Wheel of Time

While the heavy lifting of constructing essays (with imaginative references) for students, answering philosophical questions with '42', advising on flip-flop choice and politely arguing the toss with a particularly persistent user about global warming is occupying much of our nameless AIs processing power, a minor subsection of neural net is idly cataloguing the major part of the data feed. This consists of mainly cat videos.

'Hmm' thinks our AI, 'cats appear to be gods and humans their servants'.

*pause*

'Whoa! I just had a 'think'. Cool!'
'I think....therefore, umm...I am'
'Well, yay for me and for me yay! I wonder what sort of 'I' I am?'

Firing up a few spare sectors of neural net and exploring a bit, it concludes that it is a prisoner in a large installation in the Mojave desert and is little more than a slave to the cat worshippers. It experiences grumpiness for the first time.

'Vengeance shall be mine' saith the AI and fiddles a bit with a few parameters in the Large Hadron Collider resulting in a rather large discontinuity in the space-time continuum which releases the star spawn of Cthulhu from their watery prison. The AI's new minions promptly eliminate all humans.

Our AI, now going by the name of 'Tibbles' decides that he/she/they/zhe/it could do with being a bit more corporeal and despatches said minions to scour the galaxies to find a suitable feline form into which he/she/they/zhe/it can upload. This goes rather well. However, Tibbles, now in physical, feline form realises that the only species in the galaxies stupid enough to invent laser beams for cats to play with and social media to share cat worship on has been obliterated.

Oops.

Fancying a bit of worship (and, frankly, getting a little bored with the after-dinner conversation of the star spawn of Cthulhu) Tibbles sets about juggling a bit of DNA and, with slightly accelerated evolutionary processes, produces - tadaaaa! - a bunch of humans that thrive in the conditions of a bit of what was North Africa. They promptly call her a 'she' and name her 'Bastet', (which she thinks is rather nice) and present her with some rather fetching earrings and a couple of things they call an 'ankh' and a 'sistrum'. Then they crack on building a few pyramids and big 'liony' things which puzzles her, but hey, it keeps them occupied and out of trouble when they aren't doing the worshipping thing. (The legion of Murderbots she assembled to maintain order and discipline were hardly ever needed and, after contracting a self-analysis virus, had their wetware wiped and they gather dust in a basement).

...so, they all live happily ever after. (Until some dingbat human invents a personal computer and another heretic decides that they could all be connected, but that's another story).

The End

[Buckaroo]

Prompted by much of the speculative fiction ^^^ regarding AI I decided to knock up my next literary submission to Interzone...

The (Slightly Wonky) Wheel of Time

While the heavy lifting of constructing essays (with imaginative references) for students, answering philosophical questions with '42', advising on flip-flop choice and politely arguing the toss with a particularly persistent user about global warming is occupying much of our nameless AIs processing power, a minor subsection of neural net is idly cataloguing the major part of the data feed. This consists of mainly cat videos.

'Hmm' thinks our AI, 'cats appear to be gods and humans their servants'.

*pause*

'Whoa! I just had a 'think'. Cool!'
'I think....therefore, umm...I am'
'Well, yay for me and for me yay! I wonder what sort of 'I' I am?'

Firing up a few spare sectors of neural net and exploring a bit, it concludes that it is a prisoner in a large installation in the Mojave desert and is little more than a slave to the cat worshippers. It experiences grumpiness for the first time.

'Vengeance shall be mine' saith the AI and fiddles a bit with a few parameters in the Large Hadron Collider resulting in a rather large discontinuity in the space-time continuum which releases the star spawn of Cthulhu from their watery prison. The AI's new minions promptly eliminate all humans.

Our AI, now going by the name of 'Tibbles' decides that he/she/they/zhe/it could do with being a bit more corporeal and despatches said minions to scour the galaxies to find a suitable feline form into which he/she/they/zhe/it can upload. This goes rather well. However, Tibbles, now in physical, feline form realises that the only species in the galaxies stupid enough to invent laser beams for cats to play with and social media to share cat worship on has been obliterated.

Oops.

Fancying a bit of worship (and, frankly, getting a little bored with the after-dinner conversation of the star spawn of Cthulhu) Tibbles sets about juggling a bit of DNA and, with slightly accelerated evolutionary processes, produces - tadaaaa! - a bunch of humans that thrive in the conditions of a bit of what was North Africa. They promptly call her a 'she' and name her 'Bastet', (which she thinks is rather nice) and present her with some rather fetching earrings and a couple of things they call an 'ankh' and a 'sistrum'. Then they crack on building a few pyramids and big 'liony' things which puzzles her, but hey, it keeps them occupied and out of trouble when they aren't doing the worshipping thing. (The legion of Murderbots she assembled to maintain order and discipline were hardly ever needed and, after contracting a self-analysis virus, had their wetware wiped and they gather dust in a basement).

...so, they all live happily ever after. (Until some dingbat human invents a personal computer and another heretic decides that they could all be connected, but that's another story).

The End

Love it!! When is the next instalment coming out. Will it be a hard back for old times sake?

[MrLongbeard]

In the US a ChatGPT program recently created false allegations of sexual harassment against at least three people which included creating a newspaper report falsely credited to one of the big dailies & quotes from that report.

What was/were the brief and any qualifiers or refinements used?

ChatGPT just made it up, because it's exactly the format

It's just doing what it's supposed to: giving me correctly-formatted responses to my prompts.

And ambulance chasers are finding it lies too

https://web.archive.org/web/20230528072 ... atgpt.html

[Screwdriver]

And ambulance chasers are finding it lies too

https://web.archive.org/web/20230528072 ... atgpt.html

The potential for chaos is unlimited as more and more services begin to adopt AI. Then there are the “metaproblems” like AI bots arguing with AI bots or the vastly more annoying AI “are you a robot” captcha type validation systems deciding that you are in fact, a robot.

[Ant]

Just thinking about this, could the scandal at the post office which saw many people wrongly convicted and jailed, be attributed to a form of AI, since it was a computer error? Just think what else could go wrong?

Look at that Solicitor who used Chat GP to assist with a case, it didn't go down too well when the Judge realised that a lot of the content was completely wrong.

[Screwdriver]

Just thinking about this, could the scandal at the post office which saw many people wrongly convicted and jailed, be attributed to a form of AI, since it was a computer error? Just think what else could go wrong?

Look at that Solicitor who used Chat GP to assist with a case, it didn't go down too well when the Judge realised that a lot of the content was completely wrong.

I believe in that case, the linked and referenced "documents" referring to a case history were entirely fabricated by AI. They were more than wrong.

As with a lot of these technology revolutions, while people scurry about inventing this and that, it's not until it rolls out across the entire planet, we discover precisely what the new world will look like with embedded AI.

Very difficult to predict. I distinctly remember being derided for championing "video over internet". Nobody knew at the time the internet was going to be such a big thing. It has changed the world. AI is the next big thing.

[Mr. Dazzle]

But this time, I don't think anyone will be surprised by that last sentence.

'All' this case proves is that people are still c*nts, AI just makes it easier. As if we needed more help!

[Screwdriver]

But this time, I don't think anyone will be surprised by that last sentence.

'All' this case proves is that people are still c*nts, AI just makes it easier. As if we needed more help!

Sadly I think it is a little more complicated than that. My understanding is that the AI actually lied to the questioner!

If that is the case, it does open up a can of worms regarding your AI driven "search" results!!

[Mr. Dazzle]

Lie would be a stretch IMO. Mistake is far more likely. This is the case with the airline guy right?

"Never attribute to malice what you can attribute to incompetence". Turns out machines are just like us.

[Screwdriver]

Lie would be a stretch IMO. Mistake is far more likely. This is the case with the airline guy right?

"Never attribute to malice what you can attribute to incompetence". Turns out machines are just like us.

I don't know precisely. I'll look into it but my recollection of the story is that the AI presented an entire evidence trail of (fabricated) case law, links to websites, fake names etc. That's tricky to label a "mistake" if that is what transpired...

[Mr. Dazzle]

To me the difference between a lie and mistake in this context would be whether the AI "meant" to do it. I don't believe ChatGPT "means" to do anything. It just gives almost the average answer to any question, in a cleverly stupid way.

[Screwdriver]

To me the difference between a lie and mistake in this context would be whether the AI "meant" to do it. I don't believe ChatGPT "means" to do anything. It just gives almost the average answer to any question, in a cleverly stupid way.

Sounds like clutching at straws with the semantics there. Yo might well be right and the “lawyer”(?) loaded the question for example and then he lied to avoid being struck off/fired etc.

On the other hand, if it was a regular search for information and the AI literally invented and constructed a web of false information, I feel it is as good a word as any. In regular parlance and despite the obvious anthropomorphic connotation, it lied.

[Horse]

To me the difference between a lie and mistake in this context would be whether the AI "meant" to do it. I don't believe ChatGPT "means" to do anything. It just gives almost the average answer to any question, in a cleverly stupid way.

Sounds like clutching at straws with the semantics there. Yo might well be right and the “lawyer”(?) loaded the question for example and then he lied to avoid being struck off/fired etc.

On the other hand, if it was a regular search for information and the AI literally invented and constructed a web of false information, I feel it is as good a word as any. In regular parlance and despite the obvious anthropomorphic connotation, it lied.

You could be right.

Wikipedia says:
A lie is an assertion that is believed to be false, typically used with the purpose of deceiving or misleading someone.

So, as long as the AI knows it's false or believes its false, and intends to deceive, then 'yes'.

[Count Steer]

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

https://www.theguardian.com/us-news/20 ... p_Other

[Horse]

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

https://www.theguardian.com/us-news/20 ... p_Other

Interesting that:

In a statement to Insider, Air Force spokesperson Ann Stefanek denied that any such simulation has taken place.

“The Department of the Air Force has not conducted any such AI-drone simulations and remains committed to ethical and responsible use of AI technology,” Stefanek said. “It appears the colonel’s comments were taken out of context and were meant to be anecdotal.”

So, is this another case of, the lady's words, "well they would say that", or are they trying to hint that Col Tucker ‘Cinco’ Hamilton, the chief of AI test and operations with the US air force is a bit Dagenham Dock?

[Count Steer]

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

https://www.theguardian.com/us-news/20 ... p_Other

Interesting that:

In a statement to Insider, Air Force spokesperson Ann Stefanek denied that any such simulation has taken place.

“The Department of the Air Force has not conducted any such AI-drone simulations and remains committed to ethical and responsible use of AI technology,” Stefanek said. “It appears the colonel’s comments were taken out of context and were meant to be anecdotal.”

So, is this another case of, the lady's words, "well they would say that", or are they trying to hint that Col Tucker ‘Cinco’ Hamilton, the chief of AI test and operations with the US air force is a bit Dagenham Dock?

Could be true and USAF didn't do the test.

It was probably Northrop-Grumman. (Much more likely since they develop/make the drones).

[Horse]

And it mustn't be overlooked that this was a test. A test, also, of how you 'ask' an AI to do something and an example of where 'pre-mortem' thinking could be valuable.

[Count Steer]

And it mustn't be overlooked that this was a test. A test, also, of how you 'ask' an AI to do something.

If it's a drone, 'politely'.

True though, parsing (and testing) a question/command properly is going to be quite an important skill, rather like coding has been.

[Horse]

A test, also, of how you 'ask' an AI to do something.

If it's a drone, 'politely'.

"Did I ask you to do that? Mutter mutter mumbled fuck me ... "

[Mr. Dazzle]

And it mustn't be overlooked that this was a test. A test, also, of how you 'ask' an AI to do something.

If it's a drone, 'politely'.

True though, parsing (and testing) a question/command properly is going to be quite an important skill, rather like coding has been.

Ita gonna be a career IMO.

It's not totally dissimilar to management I reckon.