AI: thoughts, experiences, predictions

[Monkey]

In my opinion, AI (ML in particular), is the most dangerous technology ever to be invented; even more so than nukes or atomic bombs. I already knew this many years ago when, at Uni, I was given a choice between studying Comp. Science and Software Engineering vs Comp. Science and AI. I hated the fact that people were teaching AI and so, of course, I choose to study the former of the two.

AI, in theory, could have been used solely for good purposes. I've provided tech support for an organisation for the blind and visually impaired. They use AI, such as google lens and other apps, very effectively, instead of the eyes that they don't have. My neighbour used AI to see what colour her new fence would look best painted in. These are, in my opinion, good purposes. The problem we have, of course, is that within the human race, there are people that want to abuse technology and that is where it all goes wrong.

At this point in time, the AI/ML race is much further forward than most people realise and it's continuing at a rate of knots. The price of security is the loss of privacy, which certainly does not exist any more. Anyway, I'll leave it there.

[sinewav]

the most dangerous technology ever to be invented

I mostly agree. A couple years ago I preached to some confused faces at a barbecue that AI is a greater threat than climate change. Both are dangerous for the same reasons. A rapidly warming planet destabilizes regions and will eventually lead to widespread conflict. Information weapons are powerful enough today to destabilize every country -- and there are no countermeasures. It's possible in the next couple years we reach a kind of global hysteria we can't turn back from. The Internet is basically ruined from GAI websites, and soon the be replaced with AI agents. Social media companies are already making individualized advertisements that one other than the end user will see, so it's not hard to imagine people being exposed to a completely fabricated reality controlled by malicious actors. Scary stuff. I'm trying my best to educate everyone around me.

[kyle]

At this point in time, the AI/ML race is much further forward than most people realise and it's continuing at a rate of knots. The price of security is the loss of privacy, which certainly does not exist any more. Anyway, I'll leave it there.

Yep, way ahead. People don't realize you walk into a store like walmart, you are being watched by all of their cameras, everywhere you go, and the know who you are, mostly from you purchasing something from them with a card.
If you go to a large sporting event, they have cameras at the entrance, to identify who you are, this way if you are a known criminal, you can be caught on side. This I heard about maybe 4 to 5 years ago, they only had it at a few of the entrances of an event and were finding criminals. (A person that worked at the company that made this tell me about it, so I'm not even sure if it was in the news or not)

This was well before the whole AI LLM thing people went crazy about what AI can do. I know I mentioned this before, but now it is happening, On Sunday in Austin, TX, the first vision only Self driving car service launches from Tesla, they've already been testing driverless rides, with safety operator moved to passenger seat, for almost a month. and they are still going to keep the safety operator at the start of service, but, this scalable solution will allow many people the freedom to travel that did not have the opportunity, as I have a similar version on my car.

AI can do a lot of good, if done correctly. but training must be done correctly. on the most fundamental truths. I know you guys don't really like Elon, but I think this interview is good about his thought on AI https://www.youtube.com/watch?v=H0ALAvfmdFU I know it's pretty long, and some things go off topic of AI,I still recommend watching it all, but if you can't watch it all start right around the 32 minute mark and listen for at least a few minutes when he talks about AI safety

[Monkey]

I'm trying my best to educate everyone around me.

Good luck with that. The problem that you face is that not many people will take what you are saying seriously. If you explain to people the dangers of nukes or bombs or guns or even knives, you may well get somewhere. It's kinda obvious that they are dangerous. I don't think the same can be said for AI. I mean, to people like us it is obvious but to the average person....I think they will mostly ignore you and keep on using it..even if they use it less, they will still use it.

[sinewav]

...they will still use it.

And I'll use it too, the same way I drive a car that is dangerous or use a knife that is dangerous. I just want the people around me to understand how and why these things work the way they do so they can make the right decision to use a tool or not. One thing I told a friend of mine several months ago is that the window where these tools will be "interesting and fun" is really short and they will quickly blend into the background, so the time to explore them creatively is now. But the dangerous side of AI is just another layer of the danger of disinformation which we've been dealing with for years, so I approach that topic with Socratic dialogues and asking my friends "why do you make yourself unhappy by scrolling social media all day? You realize that most of the 'people' are not real, right?"

[Monkey]

And I'll use it too, the same way I drive a car that is dangerous or use a knife that is dangerous.

the dangerous side of AI is just another layer of the danger of disinformation which we've been dealing with for years

Some AI is far more powerful and dangerous than that; the disinformation stuff is just the tip of the iceberg. The crappy AI that the average person "uses" is also tip of the iceberg stuff.

[sinewav]

...tip of the iceberg.

I imagine you are alluding to how AI powers the surveillance state, but that stuff is totally opaque. I'm not sure what you can even do about it. You can't even go "off grid". But one of the areas I'm interested in is policy and law so I try to follow and support ethical AI whenever possible. There is no way you'll have time to read this, but I really like this article on AI governance. IMO, if AI really is a global arms race then the winner won't be countries with the best researchers but the ones with the best laws (whatever those might be).

[Monkey]

IMO, if AI really is a global arms race then the winner won't be countries with the best researchers but the ones with the best laws (whatever those might be).

It would be lovely if:
1) There were good laws governing AI.
2) The people that needed to obey them actually always obeyed them.
So, I don't agree. The AI arms race is scary shit and it doesn't just involve governments.