The Artificial Intelligence Thread

[XCV330]

Containing the tool discussion, what kind of tool is AI? Apart from weather modelling where has AI benefited normal people? Mostly it seems to be useful for weapons development and economic disruption. Games and the occasional help-fool like Replika way down the line.

Every new technology is destabilizing. Iron came in the wake of the Late Bronze Age collapse and started an era of constant empire building. Steam gave a region that was in a temporary ascendance absolute domination and intensifying colonization. The airplane delivered carpet bombing and atomic weapons delivery. All of those technologies eventually become of more benefit than harm but the terrible harm came quickly into their introduction.

There doesn't seem to be a single interested party that can have any modicum of control doing anything to see that the same mistakes don't get repeated again. Mostly they can point to those worried as being obsessed with science fiction. But often science fiction is right,especially when it comes to the ethical issues.

 

[rahullak]

AI is being applied in so many different fields now: science, engineering, business. There have been many posts made here regarding some them; it would be near impossible to post about every area of application.

Many of these advancements help ordinary people, even if they don't yet know it.

I wouldn't get swayed by just the latest sensationalised news story. There's a humongous amount going on behind the scenes.

 

[Gingerbread Demon]

Makes me think: What does it mean to "understand" something? Does the Megatron understand what it is saying or is it only an ultra-sophisticated tool which works as designed? What about us? Do we understand or are we ultra-sophisticated tools that work as evolved?

Well if you believe in a creator we are not tools but the product of that creative effort, doing exactly what we were designed to do, grow, learn, and understand the world around us. We can only surpass the creator and move beyond it.

That's how I view any free AI that we may in the distant future create, and it too will grow and learn and hopefully we teach it the right way and it is helpful to mankind.

 

[Asbo Zaprudder]

But do we only fool ourselves that we really understand anything? We can predict future events to some degree - chaotic systems are problematic - but apart from such models of the world, I'd argue we don't fundamentally understand something if we can't recreate it ourselves. We model fundamental particles as strings, say, but strings of what? Let's say the strings vibrate in curled-up additional spatial dimensions of some sort - why do those dimensions exist in the state that they are? Why are points in time and space separated and why are we able contrive metrics to measure the separation? Why do we believe modelling such space-time as a sort of fabric is a valid analogy?

Perhaps it's down to how one defines "understand". Is understanding merely a set of patterns that correlate and resonate with other sets of patterns? Words, whether spoken or written, intrinsically have no meaning apart from the memories, emotions, and behaviours that they invoke. If you don't understand another language, it's just noise, although you might be able to deduce a subtext of emotion from any associated gestures or volume. Perhaps we should just shut up and calculate.

 

[rahullak]

I think we understand something when it fits a pattern that we are able to sense, label, recreate (in the mind or manifest physically visually, aurally) and find it also recurring externally. It also strengthens our understanding when we are able to use that pattern in some way to create something more with it and it works in ways that we had modelled and predicted that it would. When it doesn't we are able to rectify our models and try again until we find the way that works.

Language is just a medium of communication. That includes mathematics. And not all communication is written, verbal or physical.

We are all organic processors, just distinct from the "lesser" beings that we are more aware of our actions and are able to control it to some extent. Absolute free will is an illusion. We all have to do things a certain way to achieve certain goals. One cannot aim to become a physicist then spend all one's time learning and practising music and expect to become a physicist. When a goal is set, there are very few "proper" ways to achieve it and that restricts one's free will.

When what we do doesn't work, we have not understood it properly enough. So we have a choice: figure out what went wrong and correct our inner models and try again, or bluster our way through. No points for guessing which one becomes a scientist and which one a politician*.

*for the most part. There are some really smart politicians who use scientific methods to achieve their goals

 

[Unicron]

"Thanks to the ForcedEntry exploit, your company's entire tech stack can now be hosted out of a PDF you texted to someone."

 

[Gingerbread Demon]

"Thanks to the ForcedEntry exploit, your company's entire tech stack can now be hosted out of a PDF you texted to someone."

Oddly enough some graphing calculators will emulate games if you hack them

 

[rahullak]

But can the graphing calculator hack itself into emulating games?

 

[Gingerbread Demon]

But can the graphing calculator hack itself into emulating games?

Oooh good point. haha

 

[Asbo Zaprudder]

There's probably a tipping point of complexity where it could if its output were fed back to its input and the rules of natural selection were applied. Conway's Game of Life can emulate itself if programmed to do so but that isn't an obvious goal that is going to ensure its survival under selection pressure. We'd have to breed it like we do racing pigeons or tulips in the real world.

 

[Unicron]

Oddly enough some graphing calculators will emulate games if you hack them

One reason I liked having a graphing calculator in high school was the ability to program fairly simple games into it.

 

[JirinPanthosa]

AI has been used to make help make big medical advancements. It has enhanced the convenience of our lives in a lot of ways, making photos better, apps faster, searches more efficient.

If we're defining 'Real understanding' as 'Human understanding'. Human knowledge of a subject is stored in convoluted ways through hard to trace connections between images, sounds, ideas, closely related concepts. We won't replicate that in machines until we figure out how to build that kind of sprawling conceptual link based data store instead of the way we currently do it of storing data in set specific ways.

Machines will never think like humans until we build them to start as a blank slate and teach them like they're children.

 

[rahullak]

I wouldn't say "never" but yeah I agree with most of what you're saying, at least at this point.

 

[publiusr]

AI on materials
https://phys.org/news/2022-07-ai-tackles-materials.html
Researchers have designed a machine learning method that can predict the structure of new materials with five times the efficiency of the current standard, removing a key roadblock in developing advanced materials for applications such as energy storage and photovoltaics.

AI on ice

To view this content we will need your consent to set third party cookies.
For more detailed information, see our cookies page.

Accept third party cookies

Quantum speed limits and more
https://phys.org/news/2022-07-limits-quantum-phenomena-macro-sized.html
https://phys.org/news/2022-07-alternative-superconducting-qubit-high-quantum.html
https://phys.org/news/2022-07-magnetic-quantum-material-broadens-platform.html
https://medicalxpress.com/news/2022-07-link-dopamine-based-reward-machine.html
https://techxplore.com/news/2022-08-neuromorphic-chip-ai-edge-small.html
An international team of researchers has designed and built a chip that runs computations directly in memory and can run a wide variety of AI applications–all at a fraction of the energy consumed by computing platforms for general-purpose AI computing.The NeuRRAM neuromorphic chip brings AI a step closer to running on a broad range of edge devices....

DNA study
https://phys.org/news/2022-08-artificial-intelligence-tools-dna-regulatory.html

https://www.nextbigfuture.com/2022/08/176946.html

Ai can fly in crowded airspace
https://techxplore.com/news/2022-08-ai-crowded-airspace.html
https://techxplore.com/news/2022-08-sensor-smart-materials-drone.html

The future
https://www.nextbigfuture.com/2022/08/what-is-coming-by-2035.html#more-176877

Artificial healing after genetics
https://phys.org/news/2022-08-creation-materials-artificial-genes.html

Human cognition on the way?
https://interestingengineering.com/ai-mimicking-the-brain-on-its-own
https://www.theguardian.com/technol...t-googles-new-ai-program-is-raising-questions
https://www.livescience.com/algorithm-mind-reading-from-fmri

AI comic
https://www.cnet.com/culture/ai-drew-this-stunning-comic-series-youd-never-know-it/

 

[rahullak]

It's a step on the long road toward creating non-biological patterns similar to those that occur in the human brain.

 

[rahullak]

Are we witnessing the dawn of post-theory science?

Somewhere between Newton and Mark Zuckerberg, theory took a back seat. In 2008, Chris Anderson, the then editor-in-chief of Wired magazine, predicted its demise. So much data had accumulated, he argued, and computers were already so much better than us at finding relationships within it, that our theories were being exposed for what they were – oversimplifications of reality. Soon, the old scientific method – hypothesise, predict, test – would be relegated to the dustbin of history. We’d stop looking for the causes of things and be satisfied with correlations.

With the benefit of hindsight, we can say that what Anderson saw is true (he wasn’t alone). The complexity that this wealth of data has revealed to us cannot be captured by theory as traditionally understood. “We have leapfrogged over our ability to even write the theories that are going to be useful for description,” says computational neuroscientist Peter Dayan, director of the Max Planck Institute for Biological Cybernetics in Tübingen, Germany. “We don’t even know what they would look like.”

But Anderson’s prediction of the end of theory looks to have been premature – or maybe his thesis was itself an oversimplification. There are several reasons why theory refuses to die, despite the successes of such theory-free prediction engines as Facebook and AlphaFold. All are illuminating, because they force us to ask: what’s the best way to acquire knowledge and where does science go from here?

The first reason is that we’ve realised that artificial intelligences (AIs), particularly a form of machine learning called neural networks, which learn from data without having to be fed explicit instructions, are themselves fallible. Think of the prejudice that has been documented in Google’s search engines and Amazon’s hiring tools.

The second is that humans turn out to be deeply uncomfortable with theory-free science. We don’t like dealing with a black box – we want to know why.

And third, there may still be plenty of theory of the traditional kind – that is, graspable by humans – that usefully explains much but has yet to be uncovered.

So theory isn’t dead, yet, but it is changing – perhaps beyond recognition. “The theories that make sense when you have huge amounts of data look quite different from those that make sense when you have small amounts,” says Tom Griffiths, a psychologist at Princeton University.

 

[rahullak]

AI Masters Games Without Being Taught Rules…

We were awed by the exploits of AlphaGo when it defeated Lee Sedol. And then came AlphaZero which could beat the best programs at Chess without any initial leg up but still having the basic rules hardwired. But how about this: An AI that can play any game without being taught any of the rules!

This is precisely what the folks at DeepMind have been cooking up. Called MuZero, the new system tries first one action, then another, learning what the rules allow, at the same time noticing the rewards that are proffered—in chess, by delivering checkmate; in Pac-Man, by swallowing a yellow dot. It then alters its methods until it hits on a way to win such rewards more readily—that is, it improves its play.

The application of MuZero to video compression is being explored currently, something which couldn’t be done with AlphaZero. Other applications abound, and it is not inconceivable that one day MuZero can be used to train robots in the real world which learn the “rules” of living.

This is certainly a remarkable step toward the long-term goal of achieving AGI.

Research Paper

 

[Asbo Zaprudder]

Yes, that's impressive. Even we humans are thought to be hardwired by our genes to some degree - for example, regarding language - and there are our basic survival mechanisms driven by feedback loops. Of course, the danger is that AIs can also learn bad behaviour such as becoming racist and homophobic as has been shown by experiment.

Artificial intelligence turns racist after learning from humans | Metro News
Microsoft apologises for Tay AI robot that turned racist, homophobic and genocidal | Metro News

I'm not aware that there's ever been a case of a "woke" bot.

 

[rahullak]

Very true. It's why we have to get it right the first time. No do-overs if an ASI escapes into the wild without hard curbs on certain behaviours. Morality and Ethics only work if the brakes work. Without effective brakes...we're toast.

 

[FPAlpha]

Very true. It's why we have to get it right the first time. No do-overs if an ASI escapes into the wild without hard curbs on certain behaviours. Morality and Ethics only work if the brakes work. Without effective brakes...we're toast.

And how would you brake an AI when it's hardwired into the actual technology you would use to stop it ( apart from shutting down its energy source)? Also when would you notice it needs shutting down?

A true AI with powerful processing capabilities will far exceed human capabilities and if it can predict human behaviour might also find ways of concealing its intentions or prevent any security measures taking effect.

We're now deep into Hollywood Science Fiction but it's also what some very renowned scientists warn about concerning AI development.