To be concerned about AI

[PennineWay]

The tech industry has written to world leaders to ask for AI development to be paused.

There are potential risks to society, economy and national security.

AI development is moving fast - faster than governments and leaders are likely to do anything about it or put anything in place. We probably can't stop this from happening at this point - the flood gates are open.

I have very intelligent friends (e.g. with relevant PhDs) who feel that the next 5 years are going to be unlike anything we have seen before, and as humans we are woefully unprepared for what might happen.

I'm concerned and think world leaders should be discussing this and urgently prioritising it. AIBU? Any thoughts?

Job losses I suppose. Leading to poverty. Social inequality between people whose jobs have vanished and those who still have work. Cyber-security? Disinformation, false information, etc. Inability to verify what’s real and what isn’t.

I'm not worried because there is nothing I can do about it.

@Bananah oh that's really interesting.

Having read the full thread yet.
just hopping on to share this video. I know the vast majority of us aren’t social media influencers but here is just a snapshot of where AI is already able to be exploited.
mots onky going to get worse.

So hitherto the main benefit of computers has been speed. They can just run iterations much faster than people and come up with a solution but how they're doing it is totally known- really basic example - excel can balance a circular reference in a spreadsheet by just running every possible solution until one works and it takes seconds . You could do that yourself but it would take years. AI is different from this because the machine is learning from experience and sort of drawing its own conclusions - not just if x then y. As a result their output is far less predictable. AI specialists admit that they already don't know how AI bots reached the conclusions/ recommended the actions that they did, but it is based on them being fed more data than you or I could experience and absorb in 100,000 lifetimes.

Another example of the difference - the early chess programmes worked by just coding every possible move and the computer being able to run through all the iterations and work out a winning set of moves given every move the opposition could make. AI works by just showing the computer loads of chess games and the computer works it out and can win every time. The output is not traceable.

Now imagine that applied to warfare or financial systems or agricultural systems. You give the computer a task, effectively, and it works out what to do based on its "experience" that has been fed to it. But that experience is determined by humans. It's also likely to be highly nationalistic.

The risk is effectively an AI arms race whereby everyone knows AI is a threat but no-one is prepared to let go of theirs because they know that the "bad guys" have it.

A trivial example, but I'm in an architecture group on facebook where people share pictures of buildings they like or don't like. Occasionally someone posts an AI-generated picture, and people HATE that - no one likes being lied to. At the moment I can generally spot them (too weird and wonderful), but while you get some people condemning the poster for trying to trick people with AI, you get others saying it's a beautfiul building, how amazing, etc.

Yeah, deep fakes are one concern but selective feeds are another. So if all your "experience" is people of Arab appearance doing violent things., you will draw a certain conclusion and certain actions (ditto white people doing violent things/ dogs doing violent things). So AI, which if fed purely objective information could be the saviour of mankind, is likely to be used for partisan ends

If you aren't worried about AI, then maybe watch a few episodes of Black Mirror.

Try Metalhead to start with - it's on Netflix Season 4 Episode 5
https://www.imdb.com/title/tt5710984/

Now imagine that as AI. Something created by humans for legitimate reasons say to win a war but has a single task to kill humans. It can't be switched off. It is almost impossible to defeat.

This is just a weaponised killer but imagine that multiplied across society. AI doing jobs, taking over digital libraries so that you have no idea what is real Shakespeare or isn't unless you physically go and pick up a book, controlling digital bank accounts with the ability to switch off the access to money anyone it sees as anti-AI or anti-state.

Take the Coronation for example. Anti-monarchy protestors were being arrested. Say there is a system set up by the state where anyone perceived by the AI to be anti state is 'controlled'. Once it is everywhere embedded into state computer systems dealing with benefits and bank accounts and passports and driving licences, your life could be just shut down. No money to buy food. No passport to leave the country.

See the video I posted just above your post - 'The AI dilemma' - if you watch this you will understand more.

You can learn about it and raise awareness. Get people talking and thinking about it. That is what is needed right now (hence why I started this thread). If you talk to one person about it you have made a difference. We all need to wake up to this.

Well, the thing about AI is that it doesn't always tell the truth. It's knowledge/ understanding is based on what is currently available on the internet, and its programming is basically to respond with what it thinks you want to hear and behave how it thinks you want it to - not to give you facts.

It could lead to overwhelming amounts of misinformation, deep fakes, images that are not real and 'information' that is not real being spewed all over the internet (and if you think it's bad now, give it another 5 years).

The internet will basically become completely useless because you will not know whether anything you read is real or not, or whether it has been written by a human or an AI. You may find something you want to buy on a shopping website - is it real? Who knows. It could become completely pointless to look something up online, purchase anything, or basically do anything that we take for granted right now.

This is just one possible outcome - there are a plethora of things that I don't know or understand (I'm not an expert by any means) but this is something I was told/ discussed with a friend who knows a lot about AI.

its programming is basically to respond with what it thinks you want to hear and behave how it thinks you want it to - not to give you facts.

You're talking about algorithms rather than AI though. So if I search "was the holocaust fake?" I'm likely to get YouTube / Facebook results that support that view and lead me deeper into that rabbit hole and that belief. So yes that is bad but it's not really AI.

AI is more like I feed a bot tonnes of videos that indicate that the holocaust is faked and then I set it off to determine foreign policy and give it control of the red button.

this is a good place to start for anyone who doesn’t know why AI presents such a terrifying risk to humanity:
https://waitbutwhy.com/2015/01/artificial-intelligence-revolution-1.html
And this article is from 2015 but some people have been aware for even longer. Now, we are on the brink of the precipice.

LOVE Wait but Why. Their procrastination article was the best of all time.

Yes it is true of algorithms but it's also true of AI.

The way that AI works is that it predicts the next word. It asks "what is most likely to be the next word here?" - it uses the internet to inform that, and also its own knowledge and understanding that it has developed.

It has theory of mind equivalent to a 9 year old (which nobody even realised it was developing until February this year) - which means it is as good as a 9 year old at trying to work out what you want to hear. That is what it is trying to do.

AI is becoming less and less like a 'bot' every day.

It is very black mirror..

Boston dynamics. Dogs with guns! https://www.independent.co.uk/tech/boston-dynamics-robot-spot-paintball-gun-b1806013.html Potentially more scary are drone capabilities and at the end of this slaughter bots presentation is something by a professor in the UK who talks about the dangers of it. Another Charlie Brooker concept from black mirror but I can say with certainty they are being developed.

I apologise if you work in the field, but, with respect, it's more than that "guessing the next word". Have a look at the predictions of Ray Kurtzweil and something called "the singularity", which he predicted would hit around 2035. That was before quantum computing and AI were integrated so suspect it may be much sooner.
Potentially 2029.

TSMC, in Taiwan are the pioneers of the smallest chips and technology that conventional silicone can do with 3nm technology. That is a fraction of a human hair size. In conventional silicon chips, that's probably about as small as we can get without problems with cooling and interference from some quantum effects. This is early release for this hasn't been applied to AI or any technology on the market at the moment for the population. When you consider that a quantum computer, for some tasks, depending on the number of Qubits, and the right programming, can be up to 100,000,000 times faster than any chip or computer that we currently have, then once it's being used for AI goodness knows what will happen. https://research.ibm.com/blog/ibm-quantum-roadmap We should really ensure that there is close observation and curtailment of some of the potentials of AI, like applying the three laws of Isaac Aminov. On the flipside as it's good to be a critical thinker, the benefits may not be as much as some people are advocating... https://www.theregister.com/2023/05/03/quantum_reality_check/

@RollerCoaster2020 Charlie Brooker was inspired by the Boston Dynamic dogs for that Metalhead episode

https://ew.com/tv/2017/12/29/black-mirror-metalhead-interview/

The video of spot in that link is so freaky it's too fast!

They are really creepy - there's video of one getting kicked over and it sort of doesn't fall but gets up almost immeidately. Here's a complilation of how they react to being kicked etc. It's quite extraordinary.

I don't like them they creep me right out.

& this is also AI freaky. Imagine when these soldiers become embedded with general intelligence AI

One of the many things that worries me about it, is, if literally millions of pp will become redundant, in both senses of the word, what will pp DO? Both for money, but also with their time?

Especially if there is no point in being creative, as AI does it better. Will we just be, as someone has already described us as, 'useless eaters' ...?

I am worried now i have read some of these articles but like other worries i'm not sure what Im supposed to do. It all feels a bit out of my hands really. I doubt I'll influence those with the power.
Id like to prepare my children, but I cant really comprehend what for.

A couple of years ago I saw a meme with 3 tsunami waves, each progressively bigger. Smallest was Covid, then we had Brexit, then we had Climate Change - I'm not sure if the order of the first two was accurate, but anyway. AI wasn't even in the picture. It seems to have landed from nowhere.

And like the other 3, there is very little that individuals can do about it.

Yes, it's very worrying for people who may lose their jobs in vast numbers, not just financially but in terms of filling days and having goals and purpose.

YANBU OP.

My DC is about to do a degree in a subject that may well be obsolete very soon due to AI. I had paid no attention whatsoever to AI until I read this. It's bloody scary.

I am worried about deep fakes either by images/videos or audio.
How easy is it for someone to do these now?

I am worried about deep fakes either by images/videos or audio.
How easy is it for someone to do these now?

Very. There are AI image generators already so you can ask them to make you a photo of the pope in a puffer jacket for example.

https://www.cbsnews.com/news/pope-francis-puffer-jacket-fake-photos-deepfake-power-peril-of-ai/