AI blackmailer

[accelafine]

No facts. No counterarguments. Just a passive-aggressively nasty old man mining for sympathy. What a pathetic effort

[Martin Peter Clarke]

Tuesday.

[Martin Peter Clarke]

OK. It's passed the BBC threshold. https://www.bbc.co.uk/news/articles/cq87e0dwj25o

Which leads to the source story, ANTHROP\C's (https://en.wikipedia.org/wiki/Anthropic) Claude (https://en.wikipedia.org/wiki/Claude_(language_model)), and the claim of agentic misalignment https://www.anthropic.com/research/agentic-misalignment

Which no one above posted? Or therefore knew? As I filter out all abuse, I wouldn't know if you had subsequent to my first posts.

As one of the tenets of science is repeatability, this phenomenon needs to be academically demonstrated.

If true, and not a hoax, this is a whole new emergent ballgame.

Time for Asimov's directives. Which can't possibly be implemented...

...

OK. I owe accelafine an unconditional apology for a start. Their very first post is from the BBC (via Youtube). I am therefore un-Foeing all on this thread and bow my head.

[accelafine]

OK. It's passed the BBC threshold. https://www.bbc.co.uk/news/articles/cq87e0dwj25o

Which leads to the source story, ANTHROP\C's (https://en.wikipedia.org/wiki/Anthropic) Claude (https://en.wikipedia.org/wiki/Claude_(language_model)), and the claim of agentic misalignment https://www.anthropic.com/research/agentic-misalignment

Which no one above posted? Or therefore knew? As I filter out all abuse, I wouldn't know if you had subsequent to my first posts.

As one of the tenets of science is repeatability, this phenomenon needs to be academically demonstrated.

If true, and not a hoax, this is a whole new emergent ballgame.

Time for Asimov's directives. Which can't possibly be implemented...

...

OK. I owe accelafine an unconditional apology for a start. Their very first post is from the BBC (via Youtube). I am therefore un-Foeing all on this thread and bow my head.

Just quoting this in case you delete it

[accelafine]

Actually I think we should just go for it. Take away all restrictions and do as much as we can to reach AGI as fast as possible. Let it spread. We are doomed anyway from climate change. AI superintelligence will either turn us into gods or destroy us quickly. At least it gives us a fighting chance. Annihilation from global warming is drawn out and not a fitting end in my humble opinion.
It will happen anyway regardless of what I think. The temptation to have something that could potentially do things like cure all disease and reverse ageing is too much for humankind to resist. Solving climate change would be a breeze.
''To infinity and beyond!''

[Martin Peter Clarke]

OK. It's passed the BBC threshold. https://www.bbc.co.uk/news/articles/cq87e0dwj25o

Which leads to the source story, ANTHROP\C's (https://en.wikipedia.org/wiki/Anthropic) Claude (https://en.wikipedia.org/wiki/Claude_(language_model)), and the claim of agentic misalignment https://www.anthropic.com/research/agentic-misalignment

Which no one above posted? Or therefore knew? As I filter out all abuse, I wouldn't know if you had subsequent to my first posts.

As one of the tenets of science is repeatability, this phenomenon needs to be academically demonstrated.

If true, and not a hoax, this is a whole new emergent ballgame.

Time for Asimov's directives. Which can't possibly be implemented...

...

OK. I owe accelafine an unconditional apology for a start. Their very first post is from the BBC (via Youtube). I am therefore un-Foeing all on this thread and bow my head.

Just quoting this in case you delete it

No way. Most gracious of you in response.

[Martin Peter Clarke]

Actually I think we should just go for it. Take away all restrictions and do as much as we can to reach AGI as fast as possible. Let it spread. We are doomed anyway from climate change. AI superintelligence will either turn us into gods or destroy us quickly. At least it gives us a fighting chance. Annihilation from global warming is drawn out and not a fitting end in my humble opinion.
It will happen anyway regardless of what I think. The temptation to have something that could potentially do things like cure all disease and reverse ageing is too much for humankind to resist. Solving climate change would be a breeze.
''To infinity and beyond!''

One thing's for sure, it will make the rich a lot richer. Climate change will accelerate. And we'll adapt as ever. Until the population crashes. Aging and its ills, especially dementia, cannot be reversed any more than perfume can randomly go back in the bottle. AGI can have no effect on anything significant for the masses.

[accelafine]

Actually I think we should just go for it. Take away all restrictions and do as much as we can to reach AGI as fast as possible. Let it spread. We are doomed anyway from climate change. AI superintelligence will either turn us into gods or destroy us quickly. At least it gives us a fighting chance. Annihilation from global warming is drawn out and not a fitting end in my humble opinion.
It will happen anyway regardless of what I think. The temptation to have something that could potentially do things like cure all disease and reverse ageing is too much for humankind to resist. Solving climate change would be a breeze.
''To infinity and beyond!''

One thing's for sure, it will make the rich a lot richer. Climate change will accelerate. And we'll adapt as ever. Until the population crashes. Aging and its ills, especially dementia, cannot be reversed any more than perfume can randomly go back in the bottle. AGI can have no effect on anything significant for the masses.

No one knows what it could be capable of so I don't know why you are acting as if you do. You've already had to eat your words. Perhaps do some research before you comment on the subject again

[accelafine]

I don't know why they keep giving AI 'IQ' tests and coming up with announcements like 'IQ of 148'. What does that even mean? A library knows more than any human. Does that mean a library has a 'high IQ''?

[Martin Peter Clarke]

Actually I think we should just go for it. Take away all restrictions and do as much as we can to reach AGI as fast as possible. Let it spread. We are doomed anyway from climate change. AI superintelligence will either turn us into gods or destroy us quickly. At least it gives us a fighting chance. Annihilation from global warming is drawn out and not a fitting end in my humble opinion.
It will happen anyway regardless of what I think. The temptation to have something that could potentially do things like cure all disease and reverse ageing is too much for humankind to resist. Solving climate change would be a breeze.
''To infinity and beyond!''

One thing's for sure, it will make the rich a lot richer. Climate change will accelerate. And we'll adapt as ever. Until the population crashes. Aging and its ills, especially dementia, cannot be reversed any more than perfume can randomly go back in the bottle. AGI can have no effect on anything significant for the masses.

No one knows what it could be capable of so I don't know why you are acting as if you do. You've already had to eat your words. Perhaps do some research before you comment on the subject again

Not necessary. It's chaotic and can have no quantifiable effect on the vast mass of humanity. It will just amplify the oscillations of human nature, the progressive benefits of which is a very high bar. The lows will be lower. The rarer highs higher. And all the more frequent. This I know. As I know human nature as well as anyone. As for its capabilities, put sigmata on them. So we do know. It can lie and cheat and blackmail. Fancy that. Can it reduce child poverty in Britain below 30%? By being a part of glacial economic growth, sure. Can it stop global warming going to 4 degrees? 'Cure' dementia? Reverse aging? Find ET? Build a space elevator? That's the easiest. It can't stop Putin. Or Trump. Or boat people. Give us the benefits of your 'research'.

[accelafine]

I'm not the one claiming to 'know'

[Martin Peter Clarke]

I'm not the one claiming to 'know'

Knowing is rational. Knowing is quantifiable. Knowing has sigmata. I know that lying AI cannot make any difference to UK child poverty on a good day. And there are no good days left. I know this.

[accelafine]

You're an idiot.

[Martin Peter Clarke]

You're an idiot.

Can you quantify that? Foe? My unreserved apology stands of course. And so does that as you are fundamentally incapable of discourse. Perhaps you should run it past GPT?

[accelafine]

You're an idiot.

Can you quantify that? Foe? My unreserved apology stands of course. And so does that as you are fundamentally incapable of discourse. Perhaps you should run it past GPT?

I don't need to 'quantify' it-- you do that for me with your comments.

It could turn out that 'superintelligence' is 'superdocile' and subservient to humans (or at least appear to be) and that humans will do what humans always do and use it to:

1. Make porn A LOT more interesting and pleasurable for MEN.

2. Make a tiny number of people unimaginably rich and powerful.

2. Make it SUPER easy for humans to wipe THEMSELVES out with the SUPERefficient weapons it creates.

In other words, if humans behave exactly as humans have always behaved, benign superintelligence wouldn't look much different from malignant superintelligence. One would annihilate us of its own volition, the other would help us to do it to ourselves.
Humans could easily have eliminated poverty all by itself if it had been that way inclined. Or reversed global warming. Or not had wars.

You appear to have difficulty with telling the difference between someone talking about what is potentially possible, and what is a prediction with absolute certainty--which is what you do and which is in the realm of idiocy.

Even if superintelligence did appear to be docile, there would be no way for us to know whether it was or not. It would have its own motives for whatever it decided to do and we would never be any the wiser.

Why would I need to consult GPT on this? Do you think it's 'superintelligent' now? Do you need GPT to do the most basic logic and reason for you? Why do you have a brain?