Is chat GPT a danger to humanity?

[Horse]

a very human tendancy to make emotional decisions concerning non-emotional systems...

Nick Leeson?

[Screwdriver]

The stock market strikes me as something AI would be good at and people bad. Lots of moving parts, lots of data, a very human tendancy to make emotional decisions concerning non-emotional systems...

We already have laws in place to limit and slow down how quickly computers can make and act on decisions in the market.

Oh yes. Yes it would and probably already is too but it's not just as simple as slowing down trades. These are vastly, unimaginably powerful intelligent systems who will eventually find a way to game the system, no matter what rules and safeguards are put in its way. It will find flaws and exploit them.

I imagine we are already way beyond the point where anyone really "knows" how these AI systems reach a particular conclusion and derive a particular answer or response. Largely (I think) because they can run a million (billion??) iterations to test a response before you could even blink! If we don't know how or why a system operates the way it does, then it shouldn't be let loose an anything so fundamentally important to human civilisation as our capitalist economy.

I think that is the crux of the matter because while it might be extremely advantageous or personally financially rewarding for those who are pushing the buttons, the net effect on the global economy could be devastating. That is my seat of the pants opinion rather than anything I could "prove'.

[Mr. Dazzle]

Could be of huge benefit too though.

Unforeseen repercussions are the hallmark of big decisions in big systems. Imagine what things would be like if you could forsee them.

I'm aware that's an argument both for and against AI

[Horse]

Financial institutions seemingly didn't notice increasing life expectancy and implications for pensions. That's people for you

[Screwdriver]

Could be of huge benefit too though.

Unforeseen repercussions are the hallmark of big decisions in big systems. Imagine what things would be like if you could forsee them.

I'm aware that's an argument both for and against AI

No doubt it will deliver a huge amount of positive benefits. I am currently playing with a text to 3d model dataset/model (?? not that familiar with the language yet. I suspect some of these early iterations are somewhat non-trivial to use for the lay person.)

Question is, will the positives outweigh the negative.

It's one of those risk/reward conundrums. Some of the rewards look promising but no one really knows what the potential risks are and there's a very good chance we won't find out until it's too late.


It's probably me being overcautious (not just me!) but the current rate of progress is astounding. These AI models are becoming more powerful every day at an extraordinary rate, way faster than anyone predicted and are also developing in ways that no one has predicted. We had a moratorium on manipulating human DNA (no doubt various labs are experimenting for "research" purposes...) but there's nothing stopping these tech geeks who are being funded handsomely by the richest corporations in the world from playing with fire.

There will only ever be one AI revolution for humanity and we're heading into it right now with our eyes wide shut.

[weeksy]

I thought it was going to take over the world by now?

[Screwdriver]

I thought it was going to take over the world by now?

It will certainly be a watershed moment in human history. The age of A.I.

No one knows for sure if there is a particular point at which an AI becomes sentient or even if that ever happens. I assume if it just keeps getting more and more powerful, eventually it will cross a line. I rather suspect no one knows when that happens or even what "the line" looks like.

One thing's for sure, we'll all know it when it happens.

[Horse]

I thought it was going to take over the world by now?

Perhaps car makers should set it to work on self-driving cars? They're struggling to do it the human way.

[weeksy]

I thought it was going to take over the world by now?

It will certainly be a watershed moment in human history. The age of A.I.

No one knows for sure if there is a particular point at which an AI becomes sentient or even if that ever happens. I assume if it just keeps getting more and more powerful, eventually it will cross a line. I rather suspect no one knows when that happens or even what "the line" looks like.

One thing's for sure, we'll all know it when it happens.

So it could be 200 years? You implication earlier was that when it's turned on it'll be one sentient and own the world. Are you now moving on slightly? Or it still could, but may be a week, or nay be 10000 years?

[ChrisW]

The stock market strikes me as something AI would be good at and people bad. Lots of moving parts, lots of data, a very human tendancy to make emotional decisions concerning non-emotional systems...

We already have laws in place to limit and slow down how quickly computers can make and act on decisions in the market.

There's a Robert Harris book - The Fear Index - with this exact premise. It was an ok summer holiday read.

[Screwdriver]

I thought it was going to take over the world by now?

It will certainly be a watershed moment in human history. The age of A.I.

No one knows for sure if there is a particular point at which an AI becomes sentient or even if that ever happens. I assume if it just keeps getting more and more powerful, eventually it will cross a line. I rather suspect no one knows when that happens or even what "the line" looks like.

One thing's for sure, we'll all know it when it happens.

So it could be 200 years? You implication earlier was that when it's turned on it'll be one sentient and own the world. Are you now moving on slightly? Or it still could, but may be a week, or nay be 10000 years?

I know nothing of course, not even that much about A.I. I find it interesting and I can see how it has the potential to have a massive effect on our technologically obsessed capitalist economy.

If you're asking me to "put a number on it" then yes, I would say it looks to me as if that watershed moment will occur sometime later this year. But then I have a slightly different perspective than most.

I don't happen to believe that "intelligence" is either rare or special but instead it is a fundamental property of our universe. As if behaviours and our observation of such behaviour reveal a deeper structural code embedded within the confines of lit. "energy" (whatever that is!) and the inevitable rise in entropy whenever and however the geometry of nature changes. Those changes are what we might refer to on a meta level as intelligent.

Self-awareness is a mere human construct which we ascribe to our behaviour because we are "special" and have developed an incredibly rich language to label ourselves as such.

The key issue, therefore is language itself and these artificial intelligences are basically language engines. Humans spent 200,000 years refining and evolving our core intelligence/language. We were able to do so because of some <unknown> evolutionary pressure which made our brain structure vastly more complex than most other animals. Nevertheless brain matter is just "stuff". A neural network of stuff out of which we developed self-awareness and the words to describe it.

These supermassive computers are themselves becoming more and more complex by the day. AI has taken a giant leap forwards, missing out on the millions of years needed for some random evolutionary mutation to create a sufficiently complex neural network - because we have gifted it to them. I **think** there may be some time before we "cross the line" and the literal complexity of these man made computer structures begin to even approach the capacity of the human brain but I don't know. What I do know is, nobody knows!

So we are left with the 200,000 years it took for the human brain to metastasise and form complex behaviours and descriptive language which we refer to as self-awareness. If we say a human "generation" is of the order of 20-30 years before our complex DNA is restructured via evolutionary pressures, then 200,000 years represents in the order of ten thousand iterations of "generative design". If you want to say the human brain took 2million years from dumb animal to self awareness, that is merely one order of magnitude longer.

You can see where I am going with this. Nothing "special" about that lump of grey matter which makes it stand apart from any other physical structure except for its extreme complexity (with apologies to our more religiously inclined, that is my position). Nothing special about the process by which a learning neural network develops and with generative design running at vastly higher rates than our relatively pedestrian 20-30 years per iteration, I fully expect one of these monster machines to be declared a a self aware conscious entity before the end of this year. And if one of them is, they all are...

[Screwdriver]

Well, it's starting to kick off...



Seems like it's not just me.

[Mussels]

Sour grapes because he lost, Google's been trying to be the leader in AI for years and now he decides it's a bad thing.

[Screwdriver]

Sour grapes because he lost, Google's been trying to be the leader in AI for years and now he decides it's a bad thing.

Lost what??

[weeksy]

These supermassive computers are themselves becoming more and more complex by the day

If that were the case then in a week they'd own the world, but that would of course have been years ago, as surely this has been the case for years already

I think you're completely wrong on this.

[Horse]

Sour grapes because he lost, Google's been trying to be the leader in AI for years and now he decides it's a bad thing.

Lost what??

https://www.zdnet.com/article/best-ai-chatbot/

The new Bing
The best AI chatbot overall

ChatGPT
The best AI chatbot if you want to experience the original

Jasper
The best AI chatbot for businesses and marketers

YouChat
The best ChatGPT alternative

Chatsonic by Writesonic
The best AI chatbot for news content creators

Socratic by Google
The best AI chatbot for kids and students

[Screwdriver]

These supermassive computers are themselves becoming more and more complex by the day

If that were the case then in a week they'd own the world, but that would of course have been years ago, as surely this has been the case for years already

I think you're completely wrong on this.

How would an AI do that?

"Years ago", we were nowhere near the extraordinary capacity for these supercomputers. Just running ChatGPT for a world audience requires staggering amounts of energy and cooling. According to Musk, these data centres are "unmissable" from space because the heat generated lights them up like a beacon.

While certainly one of the bottlenecks is sheer computing power, another is how long it takes concrete to cure! That puts a physical limit on the rate of expansion for the data centres but in some ways also highlights just how quickly the technology driving it is developing.

So far as I am aware, the race really only "started" with ChatGPT3 (and its ilk) but now it's more like an arms race; the rich and powerful are rushing headlong into this battle for world dominance, which, I think you know my opinion by now, is what this is all about.

They actually don't care what the ramifications may be for unleashing a potentially sentient AI on the world. All they care about is being first and getting a massive payback. The payback might not be of the type they were expecting.

It's all very exciting. Might be amazing, might be a disaster. Only one thing is for sure, it will be something...

[weeksy]

They actually don't care what the ramifications may be for unleashing a potentially sentient AI on the world

Based upon?

[Screwdriver]

They actually don't care what the ramifications may be for unleashing a potentially sentient AI on the world

Based upon?

Core capitalist principles; Google, eBay, Amazon (especially), BlackRock, Tescos, YouTube, CNN, MSNBC, Joe Biden...

None of them are in this game for the benefit of humanity. They're in it for the money. They will do whatever reaps the greatest financial rewards.

This is a goldrush that makes Klondike look like pocket money. No kidding, the first person to develop a <sufficiently powerful> AI will literally rule the world. You won't need "Google" your personal AI will find you the exact answer you want (or at least the one Google/Alphabet want you to believe). Same with eBay:
"Hi Skynet(!), find me the cheapest 1994 FireBlade within say 50 miles".
"Write me a decent CV and send it to the ten most likely companies oh and give me a recipe for Kima and order up the ingredients from the cheapest possible source"
"While I'm waiting, knock up a hard core porno with Kyle Minogue when she was fit. After 2 minutes throw in Nicole Kidman (just in case)".
"Then and write up a dissertation on the current state of particle physics in the style of Screwdriver. Throw in a couple of ambiguous points so that I can argue the toss over it".
"While you're doing that, make a feature length film based on The Murderbot Diaries".
"You can play the lead character..."

As if this version of humanity wasn't crazy enough already, I am expecting mayhem. If you think it's difficult to know what is "true" now, you ain't seen nothing yet.

[weeksy]

They actually don't care what the ramifications may be for unleashing a potentially sentient AI on the world

Based upon?

Core capitalist principles; Google, eBay, Amazon (especially), BlackRock, Tescos, YouTube, CNN, MSNBC, Joe Biden...

None of them are in this game for the benefit of humanity. They're in it for the money. They will do whatever reaps the greatest financial rewards.

This is a goldrush that makes Klondike look like pocket money. No kidding, the first person to develop a <sufficiently powerful> AI will literally rule the world. You won't need "Google" your personal AI will find you the exact answer you want (or at least the one Google/Alphabet want you to believe). Same with eBay:
"Hi Skynet(!), find me the cheapest 1994 FireBlade within say 50 miles".
"Write me a decent CV and send it to the ten most likely companies oh and give me a recipe for Kima and order up the ingredients from the cheapest possible source"
"While I'm waiting, knock up a hard core porno with Kyle Minogue when she was fit. After 2 minutes throw in Nicole Kidman (just in case)".
"Then and write up a dissertation on the current state of particle physics in the style of Screwdriver. Throw in a couple of ambiguous points so that I can argue the toss over it".
"While you're doing that, make a feature length film based on The Murderbot Diaries".
"You can play the lead character..."

As if this version of humanity wasn't crazy enough already, I am expecting mayhem. If you think it's difficult to know what is "true" now, you ain't seen nothing yet.

So based upon your opinion of how you see modern society then ?