Simple and clear explanation of how AI could destroy us - and what to do about it. Please read.

[Letterleaf]

A short, simple and straightforward explanation of how we could be headed for oblivion, sooner rather than later, courtesy of AI.

https://www.theguardian.com/technology/2026/sep/14/google-deepmind-ai-warnings

The AI expert urges us to contact and pressurise our governments to enact safety measures, urgently. To do this super easily and taking about 2 minutes you can go to https://controlai.org to message your local MP. The message is prewritten if you want it to be.

If this turns into a discussion please for the love of god don’t write

’I’ve asked ChatGPT and it said it won’t kill us lol.’

’It’s like they’ve not seen the terminator, lol.’

’it’s the millenium bug all over again.’ (It’s really not, and the MB was solved by the worlds computer experts getting together and solving it anyway)

It's everywhere and unnecessarily too.

Mn we don't need fucking ai thread titles. Mn functioned fine before them. Some wanky company has sold mnhq them.

People are OBSESSED with ai. They tell it everything. They are unhinged.

Haha yes. Don’t know how I’ll afford it otherwise 😬

And why do you think they’re on the hook for such large investments?

Funny how AI companies have been working on it for years and warned by others of the risks for years, but amazingly they only want to slow down and hobble their competitors now it’s become evident that selling companies a few dollars of tokens a month isn’t going to bring in the trillions of dollars they promised their investors.

Thank you so much for this! It's really interesting and informative. Heavy going in parts as I'm a bit thick but I feel smarter for reading it 😊

I remember an episode of the West Wing when Josh said that the end of humanity could come about because somebody kept back a sample of the smallpox virus.

This could be said about a lot of viruses but we have survived pandemics before. Smallpox had a 30% fatality rate pre inoculation. The Black Death was 30-60%, the death rate dropped after that and now we can control with antibiotics. Measles is deadly in a naïve population, in the 10th century it was 'more dreaded than smallpox' according to Persian scholars. It's still the biggest childhood killer in unvaccinated populations.

What I want to know is where is this intelligent AI hiding because I struggle to get it to tell the difference between decimal places and full stops.

Why do people think AI is just ChatGPT? The AI that might cause us damage is the stuff bring developed which is not in the public domain.

@FruAashild I think the point Josh was making was that anything that’s adapted to be a weapon will have different capabilities.

So the same will be true of stuff that already exists as well as AI.

@Letterleaf I was hoping you would come back and say specifically what you were worried about. Got on the side maybe you’ve just gone away and stopped worrying about it - which would be good.

It is concerning but then just take a look at our leaders, climate breakdown etc and take your pick or just go for a walk in the sun and try and manage your mind (I need to do this often, I'm not immune especially about the climate)

Hedonism is the only answer.

It’s been covered across most national news sites so have a look at some of those articles for the differing ways in which we may be affected. The Guardian article had a number of them for you to start with.

Good luck. We're only now seeing regulations for tech companies and restrictions after 25 years of the internet. AI has developed so fast in the last 18 months, even two years ago I was barely aware of it. It will have destroyed society before politicians will act. While tech billionaires laugh all the way to the bank.

I’m confused. Did any of you actually read the whole thing. He’s not saying it’s not dangerous at all. He’s saying the tech bros are all crooks, which I’m on board with. In fact he says,

“This is not the right path, and every further step down it makes the bubble’s collapse worse, as well as multiplying the dangerous and reckless experiments these companies are capable of doing thanks to their near-unlimited access to compute.”

Also, he has very little to do with AI. He runs a PR firm. Not to do with AI. He has no more knowledge than anyone with an interest in it.

If the only refute to the bad news that you can find is an (all over the place) blog from some-one with no real connection to the industry then that tells me all I need to know.

Also, I’ve no issue if you disagree with me but try not to be aggressive. And don’t accuse me of things you know nothing about like having no ability to critical think. I have a MSc in Data Science (not AI) and do understand what I’m talking about.

Sorry, but this is absolutely not scaremongering. The threat is very real and a lot of people are very worried about it.

The basic premise is that computer chips operate by Moore's Law; that is, they keep on multiplying - think 2,4,8,16,32 etc. Except that the numbers involved are massive.

Someone needs to stop this AI nonsense.

Totally agree. There's so much bullshit about this on MN.

Yes, it is very much scaremongering. Life isn't a disaster movie, but these tech bros know how naive and gullible the general public are.

I see a lot of you angrily shouting ‘bullshit’ and ‘rubbish’ but do you have anything to back up this view? All I’ve seen is a blog from someone not involved in AI (and who doesn’t actually say what you seem to think he said!) and Trump.

Who else (credible) is arguing that this is all nonsense?

I wouldn’t disagree with you at all that tech bros are disingenuous and downright Machiavellian in some instances. However, a great number of AI scientists and experts in areas such as ethics and data science are resigning from their cushy jobs and speaking out. They don’t really stand to gain anything

It is horrendous that the elite in society, the fat greedy overpaid players can do this society.
That the minions need a vote.
It is backwards.
They use popular air head celebrities as media machines, they own the social media platforms too, bring back pen and paper and inconvenience over this BS.
Not one bit of consideration for humanity or the environment as long as they get rich, planting AI data stations on land sucking the electricity and water resources raising prices for everyone.

This sums it up

Not a royalist (would abolish them!) but the King has this morning spoken out about the imminent dangers of AI and needing to reign it in. I don’t think he wades into debates that are ‘nonsense’ and ‘bullshit’ does he.

The weapons in space are either targeting other satellites or satellites targeting things like GPS (Russia already has a cheeky little satellite disrupting UK/EU GPS). There are also "defensive" technologies being explored eg a satellite shooting down or shooing away other satellites etc targetting it or shooting down satellites from the ground.

What there isn't, yet, is any talk of satellite weapons that can attack humans or infrastructure on earth. In my opinion there is unlikely to be because of the asymmetry. It would be very expensive -costingcbillions if not trillions to create such a device. But developing the tech to take out a weapon like that would only cost millions and probably less to reuse it. E.g. Chinas ability to shoot a satellite out of orbit means there is no point spending a trillion dollars on a death satellite to have them take it out the first day. Even countries with less of a space presence could, if they were being attacked, probably take out a satellite or in the worst case trigger a Kessler cascade. Which would be horrible but the rational choice for any country

So I don't think the threat from militarisation of satellites is laser death rays or a threat to life and I don't think a killer AI could take control and end us all. But the more satellites in space and the more aggression against other countries satellites that occurs will make Kessler syndrome (where shrapnel from broken satellites hits other satellites creating a cascade effect) much more likely. And that could mean an end to satellite technology, GPS, satellite internet etc and put space exploration back by a century.

The doom mongering hides the real dangers and gets in the way of a proper discussion IMO.

I think there are threats but the focus on AI becoming sentient or randomly developing into AGI doesn't help and distracts from the real issues. An algorithm doesn't need to be that advanced to cause harm- look into how Facebook designed algorithms to boost engagement over a decade ago and the huge harms that caused. It's a real life example of the paper clip scenario but it's not paid attention to because, frankly, people like Zuckerberg don't give a shit. The harms aren't catastrophic enough (ending humanity) to mean anything to them. But a badly designed system or computer programme doesn't need to have sentience to hurt people. For a historic example the East India Company was not sentient or a complicated computer programme but as a "recursively self improving system" it did a lot of harm. Although of course by "it" you can substitute the people who made it possible who were sentient and did have free will.

Likewise LLMs will never randomly "self improve" into AGI. Any more than a "self improving toaster" will evolve into also being able to vacuum the floor. But because language is so powerful and people are already outsourcing their thinking LLMs could do a lot of harm to society and cause people to do bad things. Rocco's Basilisk isn't real. It's not even a good thought experiment it's a test for psychopaths and phychosis. But if powerful people convince themselves it's real THEY can do a lot of harm.

People aren’t saying they’ll randomly evolve into AGI. AGI is the current goal they’re working towards. That’s what the race is to.

That makes the whole thing even more ridiculous. The whole point of AI is "machine learning" and "recursive self improvement". The whole reason we need these massive datacentres and very inefficient ways of running queries is because this method will, allegedly, allow the LLMs to "learn" and become increasingly better. Thats meant to be the whole difference between AI and other types of algorithm and where the threat allegedly comes from
Yet even while talking about this it's also very obvious that, as you say, the tech bros have a goal in mind they want to steer the programmes towards through programming. It's ironic. I don't think this will lead to "AGI" it is dangerous however because the people in charge are competing with each other to create the most powerful AI/to create "super intelligence". And how they define "most powerful," is as "most destructive". And they have been taught by science fiction that the sign of an "out of control super intelligence" is something that wants to destroy life. The closer they claim to be to this goal the more money and plaudits they get. So there is a clear danger there that even if the machine is not what they say it is it can still do harm if it's designed in the image if something harmful.