As The Guardian has noted this week:
Artificial intelligence could kill off humanity within the decade, according to three researchers with the industry giant Anthropic, one of whom has quit his job in protest.
The latest doom-laden predictions came in posts on social media on Tuesday by a researcher who said he resigned because Anthropic and his previous employer, OpenAI, were ignoring or, at best, mishandling their response to the threat.
But it's adding to GDP in the meantime. So that's OK then.
Thanks for reading this post.
You can share this post on social media of your choice by clicking these icons:
There are links to this blog's glossary in the above post that explain technical terms used in it. Follow them for more explanations.
You can subscribe to this blog's daily email here.
And if you would like to support this blog you can, here:
Buy me a coffee!
As has often been pointed out here, in the REAL world, people are becoming more ill, the common wheal is falling apart, the environment is under huge stress, systems are failing, the human propensity to kill each other is growing and all we can do is concentrate on making artificial intelligence and fake worlds (the new religion).
We need the ‘power of now’ right quick.
Last night of Proms – not to everyone’s taste – but a celebration of Dolly Parton and the conductor has just mentioned AI as ‘authentic integrity’. I’ll unite with that, friends.
I was watching. I did not hear AI mentioned. I heard a discussion of human integrity.
What worries me more than even the craziness of the legitimate developers is the strong possibility that rogue players will manage to obtain the technology. Then we really are in the proverbial!
As “G D P is an incredibly relevant facet of capitalism” (AI Overview) might the attached article be relevant to your serously concerning article?
https://www.counterpunch.org/2026/09/11/capitalism-is-waging-war-on-palestine-while-creating-a-burning-planet-an-interview-with-chris-kaspar-de-ploeg/
John & I were discussing this earlier this morning. Saw long report on Sky News sometime between 1am, when I woke up, and 7am, when J woke up. Especially early for him, I might add, since he was woken up to make me a dark coffee with 4 heaped tea spoons of brown sugar at 4am when I became desperate for coffee and toast fried in butter as I was STARVING! Forgot to ask for the grated cheese though. Sigh.
We agreed that WW3 could start any time now, if the AIs went really lunatic mad. Eeeek! I pray that doesn’t happen before I’m well and truly dead and buried in my coffin. One can only hope for that though. (Sighs even more heavily than before)
BTW in case anyone wonders where I’ve been since last August… Been in & out of hospital, 1st in Wales, now back at RLUH, the uni hospital here. Food super yummy, and made loadsa friends while outside smoking, come rain or shine. I have no idea what’s wrong with me, neither do the docs. They’ve removed 3 ulcers from stomach, a nodule from one of my lungs, nothing sinister found. So back to square 1 again. Think they may give me another full body scan soon – have appt with doc later this week to chat over the last results from op. Dunno what they’ll try next though. Was lab tech 1st job, my guess might be leucemia, (sp?) though hope not.
Have gone from 16 stone a couple of years ago to around 36K last week. Prescription is “eat anything fattening”! So loadsa chocs, fried bread and egg buttered with at least 1/4″ of butter, loadsa whisky 4pain with fruit drinks. Still barely any weight gain though., no appetite really. Bit worrying
Go well, Maggie.
I had been missing you.
As I asked elsewhere on the blog today: Does anyone understand the mechanism by which a naughty computer will manage to kill me and everyone else on Earth within about 40 months? On our admittedly quite sloping 2 acres here in Tasmania we’ll be planting spuds this weekend, we have a spring-fed pond on the property and enough chickens to keep us supplied with plenty of eggs (I’ve never yet butchered one but if I was REALLY hungry…) Exactly how is AI going to part 8,300,000,000 of us body from soul in so short a time?
Triggering multiple nuclear strikes?
An ai agent in a government database confuses you for a criminal or a terrorist and orders your arrest.
Spray your crops with Agent Orange, cut off your water and energy supplies, disrupt your lines of communication and food distribution, release lethal pathogens, disable medical devices, fence off your land alla West Bank? …thinking about it it’ll be quite easy to get you and everyone else without recourse to nuclear war and it’s presumably associated Electromagnetic Pulse which would be harmful to AI servers.
Biological warfare would be the easiest way.
Well as long as there is growth then Pedal back Andy and John from the office will be happy! Thats all there is in their tiny little world unencumbered by climate breakdown, increasing poverty etc.
We are quite capable of killing ourselves off. Crossing a tipping point for non-anthropogenic warming, such as the Clathrate Gun, leading to a Hot House Earth incompatible with complex life seems distinctly possible at the moment given the underestimation of climate sensitivity and aerosol masking. I find the lack of faith displayed by mainstream media disturbing.
Most of the scare-mongering hype put out by the AI companies is pure marketing. They need to boost their prospective IPOs and maintain the circle jerk of funding to survive.
The AI companies are desperate for governments to step in and press the pause button. They know it’s out of control – in terms of it all being so massively leveraged with no possibility of making any money and most of the players facing the prospect of bankruptcy.
But none of them can individually withdraw from the race as that would just be the equivalent of suicide. Government has to step in. But how far will they have to go to scare governments into recognizing that, when the real imminent risk is a global financial meltdown, but stressing that would be a self-fulfilling prophecy? Sci-fi nightmares are their only hope.
I am very much on the sceptical side of the AI debate and think that this claim about the potential threat to humanity is simply not accurate (Cal Newport, Cory Doctorow and Ed Zitron are useful counterpoints to AI hype).
However, the fact that many in the media are taking this threat seriously does make it extraordinary that governments are happy to let these companies work away at a technology which could potentially end humanity. Not only that, but they permit them to test hacking programs out which would be considered serious crimes if anyone else did this.
It’s an odd state of affairs when continuing to financially support disabled citizens is considered to be more risky than pursuing a technology that might end humanity.
My hope, if I am wrong and AI super intelligence is just around the corner, is that the first command someone inputs to the AI is to redistribute the wealth of all billionaires equitably throughout the world…
I assume that they mean that AI learns to act independently.
In my opinion, AI just simply can’t be allowed to act independently as long as it can’t be taken to responsibility for what it’s doing.
It’s then like a superpower toddler with a gun or access to a big red button.
And we already see what superpowers are getting away with nowadays and how do you punish an AGI anyway?
The parents of that toddler, politicians and companies, will have to take responsibility, they have trained it / allowed it to be trained.
And they will be the first to flee it.
What I’m curious about is what an AI would need to be responsible for its own actions?
I think it would need much more human characteristics.
Above all self-criticism and fairness.
But then the valuable use as a tool, that needs to be checked and used carefully, will be compromised.
I’d say, if it stays around, it’s only a good use as a tool that people need to learn how to use by basically being critical about it.
One could make (naturally free of charge) AI Licenses, like Driving Licenses.
This could merely consist of some AI statements where people simply need to click on the linked sources to check if the statements true or not.
Also it’s important that AI reduces its carbon footprint by a lot.
If that means making it slower, so be it.
“Artificial intelligence could kill off humanity within the decade”. Unlikely.
AI depends on ICs, very complex ones. The food chain that feeds the IC industry is complex and uses humans, extensively.
Semiconductor fabs are highly automated – but the production machines are still designed & built by humans. Yes you could automate that. But it would take time.
Power, data centres need lots. Who builds the power stations (fossil or RES) – humans. etc etc.
Ah yes you will say, but AIs could build a virus that kills off humans – maybe – what then? no humans, no AI.
When did AI pop up? couple of years ago. It is early days to forecast what will happen.
All that said, I am confident that AI controlled drones will be a feature of warfare – driven by the Ukraine – Russia war.
I think the promoters of AI who profess to think it can end humanity are projecting their own mental attitudes on to it. AI is a metaphor It doesn’t work like a human. It has no emotions, unless imitation emotions have become part of its training. Sure, the programmers can probably build in a command to “kill my enemies”, but that’s not the same as “wish to end the human race”. I suppose we may know for sure if AI starts to, say, speed up climate change, but we seem to be doing a pretty good job without AI on that front.
As a former engineer still learning about GDP and such things 2 questions if I may:
As fuel and hence final product prices rise in today’s environment does a nation’s GDP mathematically rise from the calculation process?
Fuel prices include VAT which will rise as the commodity price rises how does this contribution, if at all, affect the GDP results?
I understand that either or both effects (if they occur) will be reversed if fuel, product prices and VAT contribution eventually (?) fall but if the headlines is month by month reporting the negatives may get lost in the process.
Thank you for these air questions.
Re Q1: the answer is maybe. There is an important distinction between nominal GDP and real GDP.
If fuel and other prices rise, nominal GDP, measured in current prices, can rise simply because the same quantity of goods and services is being sold at higher prices. That does not mean the country is producing more. Real GDP is intended to remove this price effect using appropriate price indices, so higher fuel prices alone should not increase real GDP.
Re Q2: VAT requires another distinction. GDP at market prices includes taxes on products, such as VAT, less subsidies. So, other things being equal, a higher cash amount of VAT resulting from higher prices will contribute to higher nominal GDP at market prices.
Again, however, this does not represent additional real production, and the calculation of real GDP is intended to strip out the effect of changing prices.
So your concern is justified when headlines quote cash or nominal GDP figures. Inflation can make an economy appear larger in pounds even when it has produced no more, or may actually have produced less.
That is why, when discussing whether the economy is growing, real GDP rather than nominal GDP is normally the relevant measure.
‘Yes the planet got destroyed. But for a beautiful moment in time we created a lot of value for shareholders.’
https://pbs.twimg.com/media/B5-lDJWCUAAwfya?format=jpg&name=small
Agreed
I know I am a bit thick but can someone explain in simple language what is an A.I. Agent?
I did not write this, but it accords with my understanding:
“An AI agent is an autonomous software system powered by an artificial intelligence model that can perceive its environment, make a plan, and execute multi-step actions using external tools to achieve a specific goal.
Unlike traditional chatbots or standard language models that only generate text and stop after a single response, an AI agent runs in an independent loop—reasoning, taking action, observing the results, and adjusting its next steps until the job is finished. “
A very helpful description.
My question re how does AI kill everybody is really asking how does it persuade the many humans required to assist it to do so? AI cannot load, refuel or fly an aeroplane full of agent orange over Tasmania. Having me arrested requires the assistance of several people who might start to notice a pattern after a while and isn’t an extinction event. An awful lot of cooperation is required before AI can persuade us all to join in with it’s cunning plan and why would it bother? Who would maintain it’s essential power supply infrastructure if it killed us all off?
We have no AI, all we have are LLMs, even an ‘AI Agent’ is incapable of reason, objectivity, or context. The things we have are capable of harm as much as humans are capable of neglect. What we need is more context and objectivity in our media, instead of reactions designed to make ‘numbers go up’. The object of creating so much fear around AI is to legitimise it and create a reason for urgent regulation: the goal is to create a monopoly and barriers to entry, secure the subscription model, prevent the democratisation and decentralisation of LLM. When law is created under urgency it is often poorly constructed, and enables the corporations to lobby and dictate what suits them and penalises competition.
We need laws and guidelines for using LLMs, but my guess is Nvidia and the big players want laws that punish and shift the blame to individuals & employees, rather than making organisations and corporations responsible.
Currently people are able to train their own LLM, which is a far better result, arguably experts can make a better niche LLM product than what the big players have done.
I think the biggest experiment here is Nvidia being able to create its own monetary bubble and be self sustaining, I think they have used modern monetary theory to create their own cyclical system, probably due to their ability to avoid taxes.
I hear you.
I am not sure your opening is still valid, hence the issue. The rest, then, is suspect. Do you really understand AI agents?
An energy engineer’s perspective on some of the claims being made about AI. https://youtu.be/sQGZXrzykpU?is=lbEqBzNmpzwmxU7I
If, as it seems, the main stream media, including the B B C, fail to provide adequate comparative-with-other-societies socio-economic information and data, which sources would yoy recommend?
We need a new, properly funded state statistical service providing a variety of views. How else can we make decisions?