The Guardian found an AI expert who says AI won't harm us - it will just view us as ants

[highplainsdem]

https://amp.theguardian.com/technology/2023/may/07/rise-of-artificial-intelligence-is-inevitable-but-should-not-be-feared-father-of-ai-says

The man once described as the father of artificial intelligence is breaking ranks with many of his contemporaries who are fearful of the AI arms race, saying what is coming is inevitable and we should learn to embrace it.

Prof Jürgen Schmidhuber’s work on neural networks in the 1990s was developed into language-processing models that went on to be used in technologies such as Google Translate and Apple’s Siri. The New York Times in 2016 said when AI matures it might call Schmidhuber “Dad”.

-snip-

“You cannot stop it,” says Schmidhuber, who is now the director of the King Abdullah University of Science and Technology’s AI initiative in Saudi Arabia.

-snip-

Schmidhuber believes AI will advance to the point where it surpasses human intelligence and has no interest in humans – while humans will continue to benefit and use the tools developed by AI. This is a theme Schmidhuber has discussed for years, and was once accused at a conference of “destroying the scientific method” with his assertions.

-snip-

2017 interview that one links to:

https://www.theguardian.com/technology/2017/apr/18/robot-man-artificial-intelligence-computer-milky-way

Jürgen Schmidhuber has been described as the man the first self-aware robots will recognise as their papa. The 54-year-old German scientist may have developed the algorithms that allow us to speak to our computers or get our smartphones to translate Mandarin into English, but he isn’t very keen on the idea that robots of the future will exist primarily to serve humanity.

Instead, he believes machine intelligence will soon not just match that of humans, but outstrip it, designing and building heat-resistant robots that can get much closer to the sun’s energy sources than thin-skinned Homo sapiens, and eventually colonise asteroid belts across the Milky Way with self-replicating robot factories. And Schmidhuber is the person who is trying to build their brains.

-snip-

“In the year 2050 time won’t stop, but we will have AIs who are more intelligent than we are and will see little point in getting stuck to our bit of the biosphere. They will want to move history to the next level and march out to where the resources are. In a couple of million years, they will have colonised the Milky Way.”

-snip-

But in that case won’t robots see it as more efficient to wipe out humanity altogether? “Like all scientists, highly intelligent AIs would have a fascination with the origins of life and civilisation. But this fascination will dwindle after a while, just like most people don’t understand the origin of the world nowadays. Generally speaking, our best protection will be their lack of interest in us, because most species’ biggest enemy is their own kind. They will pay about as much attention to us as we do to ants.”

-snip-

[SWBTATTReg]

1. Anyone that thinks that they know what an AI intelligence thinks is full of it. No one really knows

what the AI entity will or would think, yet. And besides, in my opinion (already expressed on prior AI DU postings), we wouldn't know that an AI entity exists until it's too late, before it's well established (again, IMHO).

[edisdead]

8. there is no AI that *thinks*

not really anyway.

it IS all programming.

[SWBTATTReg]

9. Yes, I know that, but wonders can be done w/ table updates from the real world that the entity

can draw upon as a valid selection, thus a kind of 'thinking'. Millions of entries embedded in a table of entries for possible solutions based on millions of possible solutions makes it too difficult for a human mind to comprehend.

[edisdead]

10. Have worked with databases the last 27 years of my life

MSSQL, MYSQL, Oracle, dbase….

You are right it is a sort of thinking… sort of. It just doesn’t really give me nightmares the way it does others.

[SWBTATTReg]

18. Ah, a fellow IT'er. Nice to meet you, been in since '75, I coded mostly and JCL, standards, touched

Oracle w/ one of my applications, but mostly coded, standards, JCL, execute procs, etc., taught (COBOL, PLI1, etc.), bought in the preliminary Internet into SWBT (saving money from the 4000 ATT private lines that we were renting). Had an application group under me but pretty well hated that part, I wanted strictly to deal w/ IT but in order for one to advance, one had to had people in your group.
Hopefully you have better luck (avoiding the mgmt role one gets stuck w/, deal w/ strictly IT issues, etc.). Sounds like you did, w/ the DB languages, query, etc.

[edisdead]

20. Yeah man!

You go back further than me. I did manage to escape management. I got away from it for a couple years because I just burned out a little and I took a job driving school bis because ot was sorely needed. But then I ended up becoming their IT department and developing an enterprise system of record on a LAMP stack. That has been super fun to build for a company that is a startup with a purpose.

[SWBTATTReg]

25. Nice, and what a change of pace too. One does get burned out in IT, being on call all of the time,

it was one of the biggest reasons I finally retired, I got tired of being on call 24x7. Even if you had people working for you, you would still get called (due to them not responding, a variety of reasons). That is nice that you had the experience and fun in working w/ a startup, so nice and actually is fun too. Best wishes to your continued success!

[ret5hd]

2. Well then...just don't tell AI about magnifying glasses then.

Problem solved.

[marble falls]

3. Thanks. Feeling better already ...

[Shermann]

4. When's the last time a new technology surpassed all expectations?

Technology has a track record of underperforming since the 1940's. ChatGPT took us by surprise, but we'll settle back into the more familiar pattern of overhyped disappointments soon.

[hunter]

11. Personal computers and the internet sure did.

As a kid in college playing with PDP-11s and similar machines, I didn't realize I'd be building supercomputers in the 21st century out of crap I diverted from the recycling bins, and connecting them to internet pipes capable of carrying 1080p video.

The most I've ever paid for a computer was $300. That was for a shopworn i386, a long time ago.

Open source software, especially Linux, has also surpassed my expectations. The last Windows version I used was a heavily hacked 98SE.

The first real operating system I ever used was BSD. I thought that was a wonderful thing then, and Linux was like coming home again.

ARM microprocessors were pretty wonderful too. The most common computers these days are ARM systems running Linux, although most of these systems are disguised as cell phones, tablets, and smart televisions.

The first modern car I drove regularly was a used 1984 Toyota Camry. Got nearly 300,000 miles out of that. Cars I'd owned before that were primitive in comparison. I think the most primitive was a 1965 Ford Mustang my brother and I bought when we were teens.

[Shermann]

14. Computers, smartphones, perhaps

Although Moore's Law was postulated in 1965, and we haven't outperformed that prediction. AI has consistently underperformed.

Smartphones are great but require cell towers and are not as capable as the transponders promised in Star Trek.

The Internet is a sewer.

Then we come to hoverboards...

[Renew Deal]

13. The George Forman Grill

[dchill]

5. Icarus?

[Maru Kitteh]

6. Just another version of somebody telling us what "God" wants.

It's a conceit at best.

[dweller]

7. We were warned about the replicants ...

https://m.

🫤



✌🏻

[haele]

12. That's when real Ai finally comes into existence.

The current AI is still a set of macros that use existing information in a predictive manner to answer questions or do tasks. Pretty much "fill in the blank" with the answer that is closest to what matches the question. Still GIGO. Even when it writes code, it's not creating something new, it's following the process and expected results given to it by a human operator.

Haele

[Brenda]

15. I saw that show. Cylons.

Battlestar Galactica. And The Terminator. And then there's the Borg. Just like with the climate news, it seems we're living in an era where the sci-fi movies are more like documentaries...maybe from an alternate timeline.

Who will emerge, Data or Lore?

[NickB79]

16. And we kill ants when they're pests, without a second thought

Getting in the sugar and whatnot.

So his analogy isn't exactly comforting.

[highplainsdem]

17. Nor would his vision of our superintelligent robots colonizing other star

systems be comforting to other biological species out there.

[DavidDvorkin]

19. That's believable, but it makes me wonder where the alien AIs are

It seems likely to me that there are other, more advanced civilizations in the Milky Way, probably far more advanced. They would also have developed AIs, which would then have gone on to colonize the Milky Way.

So where are they? They wouldn't be interest in us, either, but we should see signs of their great engineering projects.

[Kid Berwyn]

21. And you know how people can be with ants.

Dishwashing detergent.

[brooklynite]

22. "Self awareness" is a concept that goes far beyond where AI is today.

Not something I’m going to spend time worrying about.

[Celerity]

23. it may well come though, especially if/when quantum computing starts to explode as well

[highplainsdem]

24. That's fine. Experts in the field are worrying - and warning - about it.

While also being aware of current risks from AI.

[Renew Deal]

26. There's a similar theory about why aliens haven't made contact.

The theory is that we're not intelligent enough to deal with them.

'Zoo hypothesis' may explain why we haven't seen any space aliens

Ask your friends why scientists have failed to find extraterrestrials, and you can be sure at least one of them will offer the following answer: Humans are not worthy.

We’re flawed beings. We routinely threaten one other, not to mention other species and the environment. That doesn’t sound very civilized, and it offers a plausible explanation for the lack of alien contact. Perhaps the extraterrestrials know we’re here but don’t want to deal with us — either by communicating or by visiting.

This idea is endlessly appealing. It’s also old. In 1973, MIT radio astronomer John Ball published a paper in which he suggested that the lack of success in uncovering cosmic company wasn’t due to a lack of aliens. It was because these otherworldly sentients have agreed to a hands-off policy.

They’ve kept their distance not because we’re imperfect, but because of our right to pursue our own destiny. Diversity is something that everyone in the cosmos is assumed to value, so life-bearing worlds should be left to their own evolutionary development.


https://www.nbcnews.com/mach/science/zoo-hypothesis-may-explain-why-we-haven-t-seen-any-ncna988946

[Kaleva]

27. AI can't think outside the box. We have Florida Man.

[TheBlackAdder]

30. Florida Man thinks outside the box and into the bin.

[LudwigPastorius]

28. "(Superintelligent AIs) will pay about as much attention to us as we do to ants."

...up until the point when it decides it needs our atoms for something else.

[TheBlackAdder]

29. For AI to 'view us as ants' it would first have to become sentient, else it's just program code.

[Dave says]

32. What exactly is "sentience"?

There is an area deep in the brain that, if less than a cubic millimeter is cut, ends all evidence of consciousness. It’s an area that integrates all parts of the brain. It’s more primitive than our cortex, suggesting there’s likely a lot of sentience in (at least) mammals.

But, unless you posit a ghost in the machine, it’s all “programming” — messy, evolved from slime, but programming nevertheless.

People like David Chalmers, Sean Carrol, Mark Solms, Phil Goff, Tom Metzinger, and many others kick around a lot of ideas around why and where consciousness exists (I leaned on a materialist view, above).

I like the idea that this “programming” we experience as consciousness just happens to be using a biological substrate, but why not silicon? Or maybe all of it uses a deeper substrate at a quantum level? It’s certainly possible a highly integrated AI can be sentient. We just don’t know enough to say when the “lights” turn on.

[TheBlackAdder]

34. Code, microcode, register switching, etc. can emulate aspects of a consciouis, but can't replace it.

.

At the end of the day, it's just code. And whether it's adaptive code or static, for the most part it's linear, even on parallel systems. While the appearance of 3D though can be emulated due to its speed, most is serial as different tasks swap in and out for the CPU and I/O resources.

Not sure of that cubic millimeter thing, but the cortex is primal, performing rudimentary functions and acting almost like an I/O bus for the body, so yes, if you yank base components of any system it ceases to function properly. I doubt human conscious exists in the cortex.

.

[Dave says]

35. The Hard Problem of Consciousness

Let me correct an error: If you damage two, not one, cubic millimeters of the reticular activating system found in the parabrachial region of the brain stem, it's lights out (see around the 40 minute mark):

There is no more sentience of any kind. The interesting thing about this region is it is where the signaling from all parts of the brain merge. It is where information is highly integrated. Perhaps when information is highly integrated, we have consciousness (see Giulio Tononi).

Note that the cortex can process information, make perceptions and judgements about those perceptions, all without awareness. It is only when the reticular system in the brain stem is activated do we have the light of consciousness. This answers the question of "where" consciousness resides, but not why nor how. We land in the world of David Chalmer's "hard problem of consciousness". Why awareness at all? All these functions can take place "in the dark", so to speak, so why are the lights on? No one has successfully answered this question in the thousands of years of wrestling with it.

Daniel Dennett postulates that if you can assemble a system with the equivalent of all the neurons, synaptic connections, structures, chemical swells of neurotransmitters, etc., of the brain, you then have consciousness. By analogy, Dennett says, if you assemble all the parts of a television and stimulate it's circuitry, you get an image on the screen. There is no magical image god, no "ghost in the machine", it's just the physical outcome of the assemblage. (He's a strict materialist - I am not).

So back to the idea in the OP (roughly), when does an AI system become conscious? When we assemble the equivalent complexity of at least a brain stem, anything from an alligator's or a human's, the system just might be conscious, but we just don't know. Per above, we won't know until and if Chalmer's "hard problem" is solved. So far, we are in the dark on this question.


On edit: Added the time mark for the video.

[Whiskeytide]

31. And if it decides we're in its kitchen... ? n/t

[edisdead]

33. that is an absurd statement to even make.

Which AI. What we are calling AI isn’t a monolithic thing. There are several “things” that people are calling AI…. However, I wills ay that what we are calling AI is sort of like the toy that kids are calling a hoverboard. It isn’t it is a platform with wheels. Just as the “AI” that we have available doesn’t really “Think”.

So given it is basically complex programming why would anyone venture to guess who programmed what into it and what it can be used for.