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?

As well as text, I'm worried about AI photos. Photoshop, and it's more manual predecessors, have been around for decades, but it is so easy now for fake photos to be created. There are still some glitches, but they will get ironed out and make it very difficult to spot fakes.

I'm starting to think that the only jobs that will be AI proof in the future are the physical jobs that can only be done by humans. Things like building, hairdressing, cooking, caring, beauticians, electriciams, plumbers, farming etc. Jobs that have been derided as blue collar, may end up becoming the most sought after, prestigious jobs, as they can't be replaced by a computer!

Not worried. Its a tool that we can use.

Haha, It's already here but not in an AI sense.
https://ukdefencejournal.org.uk/what-is-skynet-a-look-at-britains-military-communications-satellites/
Skynet is a family of commercially managed military communications satellites operated on behalf of the Ministry of Defence, which provide strategic communication services to the British Armed Forces and UK allies.

Skynet 5 is most recent generation of the family, replacing the existing Skynet 4 Stage 2 system. The satellites have been contracted via PFI to a partnership between Paradigm Secure Communications and EADS Astrium, parts of Airbus. EADS Astrium were responsible for the build and delivery of Skynet 5 satellites in orbit, whilst subsidiary company Paradigm will be responsible for provision of service to the MoD.

My concern is the speed it's growing and the nation's who have got their feelers out in many different international tech industries, local authorities, health sectors etc and could, with the flip of a switch, cause absolute chaos. China, I'm looking at you for this.

What makes you think that?

Basically anyone who works full time on a computer all day is at risk.
Accountants are at the top of the list due to the repetitive nature of the work. People will say AI can't provide objective advice etc but it soon will be able to.
It is quite depressing, having obtained a professional qualification in a "safe" subject.

None of it will be used to help us, it will be used to control and manipulate us.

Right ok. Then how are these governments going to deal with literally tens of millions of jobless, bank account-less credit-less, bored, unhappy, very angry people?

Jeez, scary

YANBU.

AI attempts appear to have now reached a tipping point where top people in the industry are seriously considering whether the self-learning code they set free is 'intelligent' and what that might mean.

I've been vaguely following AI attempts since the 90's and this is new.

A letter from some of them urgently requesting a pause is very alarming, not least because we know that not all countries will pause.

https://podcasts.apple.com/us/podcast/ai-with-sam-altman-the-end-of-the-world-or-the-dawn-of-a-new-one/id1570872415?i=1000610848256

i think that’s exactly the point. Just as governments have had no idea how to cope with the impact of social media, algorithms that reinforce existing beliefs into ever smaller echo chambers and the constant spread of fake news and imagery, they haven’t the faintest idea of what they would do if hundreds of thousand or even millions of people suddenly were unemployable AI

Thanks @RachelGreep87.

I agree repetitive aspects of accounting absolutely can be automated and process standardised such as processing invoices and receipts, and perhaps even some month end routines. However things like the analytical review, strategic decision making, understanding of the real-life implications of financial transactions still need humans.

I'd be surprised (or maybe I'm just naive), if all of the above could be delivered by AI in the next 5 years.

Very interesting article in the Guardian over the weekend about the dangers of AI. It's an interview with Geoffrey Hinton and I feel like it clarified some of the dangers for me.

This quote from that article was rather alarming...

"You need to imagine something more intelligent than us by the same difference that we’re more intelligent than a frog. And it’s going to learn from the web, it’s going to have read every single book that’s ever been written on how to manipulate people, and also seen it in practice.”

Hmm. I was reading about serena Williams husband Reddit Alexis Ohanian view that AI has changed so much, his whole view of things, because enormous progress in the last 5 years.

But, ChatGPT is not even that good, yet. It doesn't search the whole internet like Google does. So it comes up with things factually incorrect. Plus it only goes up to information up to 2021.

Another major limitation is that ChatGPT's data is limited up to 2021. The chatbot does not have an awareness of events or news that have occurred since then. Therefore, some prompts will render no results, such as "Who won the World Cup in 2022?" Lastly, ChatGPT does not provide sources for its responses.

So, it has its limitations. At the moment. It will develop a lot in the next 5 years though.

If you're not worried about AI then you don't understand the implications of it. It's not just about it writing essays or creating artwork. It poses a real, potential, threat to humanity if not handled well. If you don't believe me then learn more about it.

The limitations you are talking about are more likely to be overcome in the next 1-2 months than the next 5 years.

The speed at which it is developing is quite shocking.

There is already a new version which has up-to-date information.

@Minfilia @Supersimkin2

You both asked for more specific details on the concerns about AI.

For this, I really recommend a video which has already been posted on this thread. It's an hour long so watch it with a cup of tea when you have time to focus on it. I watched it in 2 parts.

It's a very worthwhile video and explains things much better than I can.

I remember a thread in early 2020 when some posters were very concerned about a virus in China, plenty of posters came along to scoff and tell them they were being ridiculous and that it would never impact us.
Deja vu

I agree. I remember someone describing it as watching with horror the approaching tsunami coming from China. MN has been pretty good in alerting pp to things that pp aren't generally aware of, the threat to women's rights from the transgender ideology was predicted on here from 2016. Re AI, I just don't know what we can actually DO about it, as it's starting to seem inevitable, and I have no idea how we can prepare for it, or mitigate it, and I don't think Governments are prepared either, look how long it took for 'revenge porn' to be classified as an offence, for example.

I'm not generally a worrier, but I do have some concerns about AI and the impact it will have on my kids' generation.

Can anyone explain like I’m five what the risks from AI would be? I understand possible plagiarism in academics and job obsoletion in some areas but these don’t seem like major risks which need to be halted.

What are the potential really worrying risks?

I understand very little of this conversation or what AI is exactly.
But how would it make the Internet irrelevant?

For example, if you google Edward Hopper it isn’t just bringing up pictures of his paintings any more - it’s also bringing up pictures of AI generated pictures in the style of Hopper. So we can’t know what’s real any more, or what’s AI generated.