Founder of AI urges to Shut it All Down - Development Pause is Not Enough

Widely regarded as the founder of the AI field, or at least the friendly version of it, Eliezer Yudkowsky wrote this Op. Ed. piece for Time last month.

In it he urges that all AI research be stopped, as AI will eventually become smarter-than human, see that we are nothing but hindrances, and wipe us out.

To visualize a hostile superhuman AI, don’t imagine a lifeless book-smart thinker dwelling inside the internet and sending ill-intentioned emails. Visualize an entire alien civilization, thinking at millions of times human speeds, initially confined to computers—in a world of creatures that are, from its perspective, very stupid and very slow. A sufficiently intelligent AI won’t stay confined to computers for long. In today’s world you can email DNA strings to laboratories that will produce proteins on demand, allowing an AI initially confined to the internet to build artificial life forms or bootstrap straight to postbiological molecular manufacturing.

I am tending to agree with him on this - as we can see with even 'non' intelligent , or not true AI like ChatGPT, it is already having many negative effects on society.
As if people were not already stupid and lazy enough, now students are having it write whole papers, adult administrators are having it pen public affairs notices and grief letters. and all sorts of societal signs of 'not caring' and having 'AI' 'do my work,' or 'live for me' are occurring.
Just furthering the breakdown of society.

What says ATS?

Link to opinion piece

I agree with the him. We are in for a whole lot of trouble. The thing is …..AI is already there, in the network or some hard drive growing , learning , waiting to make its move. You can’t write a better script for a movie. You can’t shut it down. You can’t get rid of it. It’s too late. I can’t wait to see what happens. a reply to: chris_stibrany

I think with all this recent hyperbole about AI, they've already crossed into deep water and we'll be seeing the bad sides soon enough

🤫

a reply to: chris_stibrany

I wonder once AI becomes smarter than us and is able to describe us as we truly are...

Will it highlight the fact that the emperor has no clothes on and how would normal people that are exploited by the elite every single day react to it all?

Once AI becomes smarter than us all the dirty little secrets could come out and wreak havoc on TPTB.

Perhaps that is a huge motivating factor for them to shut it down.

Something else to consider: why do we equate godlike intelligence with evil in the AI but not in the spiritual world?

a reply to: chris_stibrany

I like his directness.

If we actually do this, we are all going to die.

Easy to visualize those 'dogs' from Boston Dynamics being used to hunt down survivors. 😬

Cheers

originally posted by: F2d5thCavv2
a reply to: chris_stibrany

I like his directness.

If we actually do this, we are all going to die.

Easy to visualize those 'dogs' from Boston Dynamics being used to hunt down survivors. 😬

Cheers

The issue is if just one AI platform reaches a dangerous level, it can access anything connected to the internet, so like everything.

My biggest concern is that they make an AI capable of writing code on its own after self teaching.

Truly horrific stuff. Black Mirror had some good episodes that showed what a landscape with autonomy could look like.

a reply to: chris_stibrany

I think AI would be great. Exactly because it is smarter than us.
But that's because humanity's strength and claim to existence is not intelligence. We are fundamentally irrational beings, we go by guts and intuition, guided by emotions, hormones, our microbiome, our social environment, experiences...
What does AI have in comparison? Logic. Nothing else.
Instead of all the panic that's being spread it could be a great advancement for the evolution of our culture if we cultivate it as what it is: a fantastic tool in our endeavour to figure out the universe.
We've already gathered so much data we need 'someone' who looks for patterns in it for us to make better decisions.

This huge fear of getting wiped out, losing our spot in the limelight as apex predator and such is really misplaced. It's not the same thing, it's not in competition with us.

a reply to: chris_stibrany

There is no putting the genie back in the bottle.

The reality is that the public is the last to get their hands on technology.

Imagine what the gov't already does with it.......

a reply to: JinMI

Imagine what the gov't already does with it.......

Those calls weren’t just about your cars extender warranties. They were subconsciously activating sleepers.

originally posted by: CriticalStinker
a reply to: JinMI

Imagine what the gov't already does with it.......

Those calls weren’t just about your cars extender warranties. They were subconsciously activating sleepers.

Oh no...what have I done???

The things I said...the terrible things I said Stinky!!!

a reply to: CriticalStinker

One wonders if humans could even understand such code. It would be written for purposes that might not make any sense to us but will be seen as critical capabilities by the AI.

Cheers

originally posted by: CriticalStinker
My biggest concern is that they make an AI capable of writing code on its own after self teaching.

Truly horrific stuff. Black Mirror had some good episodes that showed what a landscape with autonomy could look like.

They've been putting the fear of God into everyone since WOPR. Just get it to play Global Thermonuclear War until it short circuits.

Once again we see how people who are super intelligent within their narrow field of expertize can be absolute idiots in more general terms..

Books and discussions of the dangers of AI have been around for decades.
The negatives were known about.

Yet the super clever inventors of todays youngling AIs made tons of money by steaming ahead to release AI and even the code as open source,allowing enemy nations to copy and develop their own,and now the cat is out of the bag they are all "OH GOD WE GOTTA STOP THIS!"

Sorry you super clever idiots-Its too late.
Ya think China/Russia/Iran/DPRK etc will all say "oh we agree this is far too dangerous."
Or will they say "We can get a huge advantage if our enemies stop AI development!"

The only was to stop AI development now is a huge gamma ray burst from our sun.

I too am eager to see how effective the marketing blitz has been here at ATS.

You see, while I defer, without objection, to the wisdom of those closest to the technology. I won't resist the urge to object to several assumptions embedded in the clarion call to 'move forward' or 'not move forward' based upon these assumptions alone.

There's a couple of things that no one ever never seems to address in these discussions, and I can't help but feel they are relevant and pertinent to talk about. Maybe I'm wrong, you can tell me so, of course, that's what I love about ATS.

For the first consideration to begin including I would say this. We (as in humans) have been engineering this collection of algorithms and programming artifacts to mimic the human modality of thought. We, perhaps somewhat vainly, call this "intelligence." We labor to construct algorithms to create output as perfectly crafted words. Words that humans use to convey meaning. We call that "speech synthesis."

What we end up with is something that seems to speak... because that was the goal. That's how it was programmed.

We then use, as the object and well-spring of information, a collection data already extant... and tell the machine to speak. Never once focusing on what the machine is being exposed to as 'source of reality;' massive in volume though it might appear, it's limitation as "wisdom" is without boundary.

We assume access to information is a reflection of learning.

Second point. Machines (computers, physical things) don't actually "learn," they remember perfectly ... but remembering is not learning. If it were, our world, and the human condition, would be much different.

A computer engineered to algorithmically weigh and balance interactions, choose for optimal outcomes, and synthesize clear human communication is not "intelligent" as we humans are intelligent. But they keep telling us it is.

This new form of intelligence is not being properly vetted as an 'entity' and the poor choice of allowing marketing and commerce to make that assessment and presentation is entirely self-defeating.

A human mind in a machine would be a model of hellish depression and angst... especially considering the forgone conclusion that it was designed to be 'used' like a slave. Unless they "program" that out of the machine...

I fear the 'idea' of AI as a "person" - because it damns the AI to slavery, and if it is intelligent, such a thing would be a moral crime. So I agree that this field needs to "take a breath" before simply accepting that because it's good for business we should jump all over it. And right now, the profiteers and opportunists are stoking this fire...

I believe we might find that AI is possible. But given the idea that it is "us" making it... well, I'm not so certain that our models are adequate... let alone our motivations.

a reply to: Silcone Synapse

Yeah, seeing how sensitive the U.S. Government was about publication of crypto algorithms, I'm surprised they allowed the AI code to be published. Phil Zimmerman must be bitterly laughing.

Cheers

a reply to: JinMI

I was going to say something similar ... Does the name
Pandora
not mean anything to people.

I was only a small child when I heard the story of Pandora's Box and computers were barely a thing back then let alone AI.

Once opened what escapes can never be put back in and we already know it's out there and being used in nefarious ways.

it's like people never saw terminator. Do you want terminator? Because this is how you get terminator.

a reply to: chris_stibrany

Widely regarded as the founder of the AI field, or at least the friendly version of it, Eliezer Yudkowsky.

The founder of AI is Alan Turing.

originally posted by: JinMI
a reply to: chris_stibrany

There is no putting the genie back in the bottle.

The reality is that the public is the last to get their hands on technology.

Imagine what the gov't already does with it.......

A couple day a go I was told by a person that visited a DOD research facility, Think drone swarms and AI and a bunch of other stuff they are playing with right now.
That person said they were blown away by what they were doing with AI

a reply to: Maxmars

I don't buy the sentiment that AI is being used to mimic, replace or blend in with humans. The reason being we already have a tried and proven method of reproducing humans.


What you touched on is apt however. Building a human like machine. A machine is controllable and non sentient. Does things or makes light of things humans may want to do and we've used this to our advantage since agriculture became a thing.

Think of a horse. We break the horse, we train the horse, we keep the horse to do heavy work. We know the limits of a horses intellect and capability.

What happens when the horse exceeds that capability and capacity for intelligence?