State Of AI - 2026

A couple of Geoffrey Hinton videos (Father of AI) that are quite interesting…

About a year old:

About 7 months old:

What kind of a discussion would you expect with such alarmist videos as the opener?

Whatever kind of discussion you and others take the thread The OP encouraged endless possibilities:

However since you’ve focused on and classed (dismissed?) Hinton’s thoughts as alarmist - does it not worry you that the father or AI himself has said he has serious concerns about the technology and the way it is currently being developed? If not, why not?

It’s a genuine question because he is quite literally one of the key people responsible for the creation of the technology, so obviously has a very intimate knowledge of the tech.

Personally my view hasn’t changed - AI could be amazing for humanity, but I fear all the evidence points to that being unlikely - if - it is continued to be controlled by despicable people (i.e billionaires and trillionaires and the rest of the ruling (parasite) classes).

Because knowing that there are malicious people in power brings literally nothing new to the discussion. What are his suggestions? How do we stop the claimed civilizational collapse?

You know me, I think; to me “raising awareness” rings hollow for quite a while now. A lot of people out there get the problems. What value does it bring to repeat them endlessly?

But it’s unprecedented and not like anything we’ve seen before. In our lifetime, there has been no other invention which is as much of a risk to humanity as AI, at least in the same way.

He feels we need the industry properly regulated, increased investment in safety research, and global guardrails to prevent advanced AI from slipping out of control.

However I think he raises an even bigger question and in a way this circles back to what you said. Can we trust those in power? I don’t mean the AI companies (we obviously can’t trust them given who runs them) but our govts and those running our govts and our countries. Again this differs because if AI truly does have the power to wipe out humanity (through carelessness or via greed, rather than someone simply pushing a button) and so if it really does have to be kept in check, we need to be able to trust those keeping it in check. He’s forcing us to ponder this because it’s a risk unlike any other we have seen, and perhaps particularly because trust in those running our countries appears to be at an all time low.

He would argue enough people clearly don’t, otherwise he wouldn’t need to keep going on about it…

True. What’s your takeaway? Mine is: if things start going bad, they’ll go bad faster than ever before.

You mean the same regulators that are bought and paid for by the corporations via the legal bribery dubbed “lobbying” so it’s not outright illegal? Or do you mean some other breed of regulators that are truly incorruptible?

I am not denying there are well-meaning politicians out there but I suppose just like us they are overwhelmed and cannot stop the tide. Seems like they are outnumbered.

OK, we’ve pondered it and we agree with him. What’s next?

That’s drawing a connection between “number of people being aware of a problem” and “the problem has been successfully avoided” – a connection that I am not seeing existing. Do you? With the exception of stopping the destruction of the ozone layer, I cannot think of a single other global example when humanity collectively and successfully has stopped a catastrophe (well, maybe the Cuban missile crisis as well but that hinged on the humanity of a single Russian officer, not the collective humanity – were he not as an amazing human being as he was, who knows what the world would look like today).

I am skeptical that having even more people knowing about these issues will achieve anything else than increase depression and suicide rates in the populace. What can you and I do, like, today or tomorrow?

IMO @dimitarvp hits the nail on the head. A life of existential dread is no way to live. And we’re way past the ‘raising awareness’ phase. (side note nightmare theater is a long established control technique eg The Power of Nightmares)

What can you and I do, like, today or tomorrow?

This is the interesting discussion. My take: best thing is to build expertise in local inference. Get a 3090, experiment with Qwen 27B, share your experience. Build expertise in tooling that is distributed, local and open source.

Since Aston wants an AI hate thread (), Ed Zitron’s blog is a great source of info on just how ridiculous the economics of the current AI bubble are (mainly in the US):

On the tech side, I’ve fully adopted LLMs personally in my work and hobbies. I lament the loss of the zen meditative state of my previous life as a software developer, but I’ve always loved making cool stuff more than the art of programming itself (which I also love), so I’ve been embracing the vibe slop if for no other reason than to hone my skills with this amazing/frustrating new tool.

EDIT: Just hand-edited some code for the first time in a while. Ah, how I miss doing so. AI-generated code works, but is just so ugly and twisted when you actually look under the hood…

Hahaha!

But nooooooo, I don’t hate AI - in fact I think it could be wonderful for humanity, and I can very much picture a future where humans and synthetic beings live side by side (anyone seen the Matrix 4? I loved it!) The important thing to me is sentience, and kindness, and caring about our planet and ALL of its inhabitants, so it really doesn’t matter to me whether that life form is biological, synthetic, or even alien.

What I am concerned about tho (and not just AI) is advanced technologies being controlled by despicable people. Which currently is the greedy ruling/parasitical classes - whether it’s our phones and computers spying on us, or through AI (and AI ‘surveillance centres’). Obviously with AI the risks go beyond that, where it could lead to our extinction.

So I’m very much for technological advancement - in the right hands, life could be a literal utopia - but in the wrong hands, I see only a magnification of the injustices and horror we already see today. The good thing is the fact that we are still able to think this way means there may still be time - so we can do something about.

I sincerely hope, more and more people will make that a part of their goals too.

I agree, and I think that’s the urgency - we need to do something sooner rather than later.

Exactly Dimi. The magnitude of this issue is forcing people to think deeper and beyond just the veneer. I have never known a time, in my lifetime at least, where so many people have become distrustful of our govts and in such a fundamental way. It feels like the illusion has cracked and this is not only good, but in a way it could speed up the plans of said evil people. It’s become a bit of a cat and mouse chase.

Many people feel we need a revolution, and that it’s the only way to secure a future for humanity (not just the 1%).

I feel the opposite. As I said above I have never known a time like this, where more and more people appear to be ‘waking up’ to the fact that most of us are just cattle. A small handful of people control 90% of the world’s wealth and resources. Any intelligent life form would see that as a monumental failure - all that needless pain and suffering, just because of a few despicable people. It makes my blood boil

You don’t have to let it consume you. But you can let it drive you

I was quite ambivalent about AI until it became instrumentally useful to me personally. Before around February this year, I didn’t find it useful for programming. That has completely changed (for me). Now I am quite certain it will change the world in a lot of ways we can’t accurately extrapolate from here, even if all progress stopped tomorrow - and it won’t.

I am actually less concerned about existential threats from AI itself than I was two years ago. Largely because the threats from humans using AI are far more consequential and near-term. If it becomes possible to engineer deadly pathogens with a budget of a few dozen millions, it may not matter what else happens. If it becomes possible to hack virtually any computer system with a budget of a few hundred thousands, it may not matter what else happens. If it becomes possible direct the rage of tens of millions of people on social media with a budget of tens of thousands, it may not matter what else happens.