Artificial intelligence: is it that dangerous?

[Scout]

AI is not just "new technology." It never forgets anything it learns, and it's far more intelligent than any human. If the programmer is biased, so is AI. You've heard about it being asked why one should vote for Kamala and it gives many reasons; asked why one should vote for Trump, it has nothing to say. Does this not prove that programmers have too much power? What if, for example, the programmers instill the idea that the world would be better off without humans? In any case, do you doubt that AI, as intelligent as it is, will eventually find a work-around of its programming and write its own programs? Become autonomous? We have to think long-range, not just about immediate economic consequences.

[Ian]

AI is not intelligent. It just has immense computational power and programmers can set it on a logic path. I saw one exchange outlined between someone and AI, and eventually AI gave him exactly what he wanted. In my opinion, it was rubbish. "Nature" gave it a task and got it to recycle the problem based on what it had already produced, The original got degraded.

My one and only encounter was to ask it to generate an image for my next ebook on quantum mechanics. The results were useless

[J.]

To be fair, presenting a comprehensible image of quantum events is trying to explain absolute randomness in a deterministic world.

[Ian]

To be equally fair, the Schrödinger equation that founds quantum mechanics is fully deterministic. QM is NOT random - it is deterministic with superimposed variation limited to a quantum of action that arises as if the particloe was following a classical trajectory, b ut was being bounced around by an accompanying wave.

If you think of Brownian motion, the "degree of randomness" increases with time. In quantum mechanics it does not. It is just some "jostling" as de Broglie wrote arising from the deterministic wave.

[Papaphilly]

Scout wrote: "AI is not just "new technology." It never forgets anything it learns, and it's far more intelligent than any human. If the programmer is biased, so is AI. You've heard about it being asked why one ..."

AI is not intelligent, it is programmed.

[Graeme]

Ian wrote: "I didn't say they were, but now it is much harder for such people to enter the country so I assumed that was a one-off attack. Effectively, it was the distance argument in place, but I concede slack border control could lead to another 9/11...."

The Biden/Harris open US border has enabled any number of terrorists and criminal gangs to cross with impunity.

They are already making themselves felt amongst the local population.

[Graeme]

Papaphilly wrote: "AI is not intelligent, it is programmed...."

A system only has to mimic intelligent behaviour to be indistinguishable from intelligence.

[Graeme]

Sai (the climate catastrophe is real) wrote: "Thats a true point. What I meant by that the economy would crash is that AI will start replacing workers because its cheaper. Economics is not my strong suit, I could be wrong also, but people woul..."

Guaranteed that some people will lose their jobs....

The rest of us will be fighting in the rebellion against Skynet(TM)...

[Sai :)]

Found a disturbing article online....

https://www.msn.com/en-us/news/techno...

i wish people would study the effects of things like character ai more before releasing them to the public

[Graeme]

Sai (the climate catastrophe is real) wrote: "Found a disturbing article online....

https://www.msn.com/en-us/news/techno......"

The family will be devastated.

The bottomline - human relationships are important for health.

[Ian]

Graeme wrote: "Papaphilly wrote: "AI is not intelligent, it is programmed...."

A system only has to mimic intelligent behaviour to be indistinguishable from intelligence."

Not so. It has to be able to induce something from outside its data base. That is creativity. If it restricts itself to its data base it is mere computing.

[J.]

Ian, an AI's database would easily include the sum total of all human knowledge.

[Graeme]

J. wrote: "Ian, an AI's database would easily include the sum total of all human knowledge."

Oh, so it'll be permanently confused by multiple contradictory ideas...

[J.]

Or it gets hung up on that old joke wherein scientists build the most powerful computer ever built and ask it if there's a god.

"There is now."

[Papaphilly]

J. wrote: "Or it gets hung up on that old joke wherein scientists build the most powerful computer ever built and ask it if there's a god.

"There is now.""

That is a great joke.

[Ian]

Graeme wrote: "J. wrote: "Ian, an AI's database would easily include the sum total of all human knowledge."

Oh, so it'll be permanently confused by multiple contradictory ideas..."

This happens now. What Chat GPT seems to do is discuss. the problem with the questioner to find out what is already believed, and it supports it. In most cases it will discard anything that does not agree with its answer.

It could never disprove or falsify something which had general approval. of course, in modern science, neither can a scientist in kost cases because of the herd mentality of the majority.

[Sai :)]

Graeme wrote: "Sai (the climate catastrophe is real) wrote: "Found a disturbing article online....

https://www.msn.com/en-us/news/techno......"

i completely agree

[Papaphilly]

Sai (the climate catastrophe is real) wrote: "Graeme wrote: "Sai (the climate catastrophe is real) wrote: "Found a disturbing article online....

https://www.msn.com/en-us/news/techno......"

This should be interesting. This is probably the first test for AI as a responsible party.

[J.]

Well, devising an AI voice might have spared Wildwood, NJ the current lawsuit by a woman whose recording of a very familiar phrase that plays over the loudspeaker on their boardwalk - "Watch the tram car please!" Originally recorded to caution pedestrians when the famous Wildwood boardwalk tram car's approaching, the plaintiff is claiming that her recording has been used by tourism and memorabilia for profit.

[Scout]

No one's talking about the effect the bias of programmers has on AI.

[Ian]

Or their competence, which is more dangerous

[Sai :)]

Scout wrote: "No one's talking about the effect the bias of programmers has on AI."

I agree, there is lots of bias in AI. Whenever I need to generate an image for something because I can't find what I want on the internet I have to try really hard to get the prompt right if I'm generating a picture of a girl or a woman because all the AI knows how to generate is women in low necklines and spaghetti straps because that's all its been taught, and it's really irritating and especially misogynistic because hardly any women actually dress like that.

[Ian]

If you think that is difficult for it, I recently asked it to generate an image of an atomic orbital for the cover of an ebook I am writing. Thi8s should be fairly straightforward, but it id not seem to know in any depth what quantum mechanics is about. It went off in all sorts of fanciful directions. If it can't generate an image from an equation that has been known for 100 years, what else are its failings?

[J.]

Pop its CPU. Ask it to draw a white guy.

[Graeme]

Hi Sai, I use Grok on X and it doesn't have that problem.

[Sai :)]

Graeme wrote: "Hi Sai, I use Grok on X and it doesn't have that problem."

I guess different programs have different biases then. I've never used Grok before but I'll try it out the next time I need something.

[Sai :)]

Ian wrote: "If you think that is difficult for it, I recently asked it to generate an image of an atomic orbital for the cover of an ebook I am writing. Thi8s should be fairly straightforward, but it id not se..."

That's interesting. Maybe it just didn't get fed enough data about the atomic orbitals? AI has a similar problem with going crazy when drawing hands because of the lack of data. It might be that?

[Ian]

Sai (the climate catastrophe is real) wrote: "Ian wrote: "If you think that is difficult for it, I recently asked it to generate an image of an atomic orbital for the cover of an ebook I am writing. Thi8s should be fairly straightforward, but ..."

Perhaps, but the internet has many links so it should be able to manage if it takes in what is on the interent. I think the problem is the programmers want such illustrations to be "spectacular". Reality doesn't matter, maybe.

[Scout]

I asked if bias on the part of programmers is dangerous, and Ian replied, "Or their competence, which is more dangerous." So there is danger with AI regarding programming. Given that, programmers could make mistakes or intentionally program AI to harm humans. Right?

[Sai :)]

Considering that every time a new technology comes out a new type of crime is introduced to the world, that sounds pretty accurate. I'm not against AI but if we aren't careful it could do some serious harm.

[Papaphilly]

Scout wrote: "I asked if bias on the part of programmers is dangerous, and Ian replied, "Or their competence, which is more dangerous." So there is danger with AI regarding programming. Given that, programmers c..."

Yes

[J.]

Papaphilly wrote: "AI is not intelligent, it is programmed."

Let's say your position is correct. If an AI is programmed to emulate intelligence, sentience, consciousness, and/or agency so perfectly that it is indistinguishable from a human on another computer, does your position matter?

[Barbara]

I have noticed that there are a few AI transcription services cropping up - basically you read to AI, teach it syntax, grammar, spelling (for subtitles). I don't know how legitimate they are or what is actually involved but I wonder if AI appears to be opening up a whole new job category, while shutting another one down.

[Scout]

I've noticed AI on Amazon and Google. Basically it summarizes information from reviews on Amazon and from sources on Google. I don't trust it, as I read what AI summarized about a product on Amazon and then read the actual reviews, and AI had left out information provided by negative reviews (supposedly to enhance the chances that you'd buy the product). I'll stick to reading actual reviews.

[Papaphilly]

J. wrote: "Papaphilly wrote: "AI is not intelligent, it is programmed."

Let's say your position is correct. If an AI is programmed to emulate intelligence, sentience, consciousness, and/or agency so perfectl..."

Yes it does. Assuming we actually develop an aware AI, they would have to have rights because they are self aware. If one mimics self awareness, but is not self aware, then they can be turned off without a thought. Now saying it is indistinguishable may make it tougher, EXCEPT the programmers know it is not self aware because it was programmed to be indistinguishable. If it is real self awareness, that is a game changer.

To give a different analogy, but with the same idea. If someone created a Van Gogh that fools everyone, is it a Van Gogh? No because Van Gogh did not create it. If it is a forgery, fraud, or even a homage, it is still not a real Van Gogh.

[Papaphilly]

Barbara wrote: "I have noticed that there are a few AI transcription services cropping up - basically you read to AI, teach it syntax, grammar, spelling (for subtitles). I don't know how legitimate they are or wha..."

That is the way of the world. Innovation eliminates certain job categories. It will create other jobs and very good paying jobs. Some get left behind and that is a shame, but then that has been the way it is since the very beginning.

[J.]

Google AI chatbot responds with a threatening message: "Human … Please die."
https://www.cbsnews.com/news/google-a...

[Nik]

Hmm.. menacing but polite, we’ll see where it develops

[Nik]

AI 🤖 can start feeding on banking small fry:

https://cio.economictimes.indiatimes.....

[Papaphilly]

Technology changes environments. Some will lose jobs and others will gain jobs as it changes.

[J.]

Davidson County Schools moving forward with AI surveillance program
https://myfox8.com/news/north-carolin...

[Papaphilly]

This could be good if used properly and not abused. I do worry about privacy concerns and mistakes.

[Graeme]

Nik wrote: "Hmm.. menacing but polite, we’ll see where it develops"

I definitely prefer to be exterminated by a polite robot than a crass robot...

[Graeme]

Papaphilly wrote: "This could be good if used properly and not abused. I do worry about privacy concerns and mistakes."

FADE IN.

EXT: Metropolis Central. Street Scene. Robocop #12456 encounters Citizen Max Walker.

Robocop: (accosts Max Walker) "Your advanced age of 148 years, 4 months and 3 days has been noted. You have been designated for economic euthanasia to improve the GDP. You shall comply!"
Max: (Suprised) "But ... but ... but ... I'm only 44!"
Robo: (Arches Electronic Eyebrow) "The Central AI Database does not make mistakes. You are 148 years, 4 months and 3 days old, and well past you expiry date!"
Max: (Perplexed) "But ... C'mon ... humans don't live to 148 years!"
Robo: (Determined. Lifts hyperdermic syringe filled with a green fluid) "You talk a lot for a 148 year old human. Stand still while I apply the euthanasia!"
Max (backs away) "No Way!" (Turns and runs away)
Robo: (Nonplussed) "Very speedy for a 148 year old human!"

FADE OUT.

[Nik]

An excellent demonstration, Graeme :)

[Scout]

Good one, Graeme.

[J.]

This seems appropriate.
https://youtu.be/kjd5ksm5UMA?si=4oUrO...

[J.]

A strange thought occurred to me.

Could you create an AI to emulate a specific judge? Could you pump it full of everything the judge has ever written and said, then run arguments against it until you find a perfect way to the rulings you want?

[Scout]

Oh, that's just the beginning. What scares me is that now it's a competition with China to have the best AI and it seems that's overshadowed any discussion of safety. Maybe I've just missed those discussions? It seems AI development is most important. Skeptic that I am, maybe we're creating a monster?

[J.]

Artificial intelligence used to locate rare Aussie bird for first time in more than 30 years
https://www.9news.com.au/national/pla...