Anthropic CEO frets about 20% unemployment from AI, but economists are doubtful Anthropic CEO Dario Amodei is worried that AI could eliminate half of entry-level white collar jobs in five years. There are many other potential labor risks lurking about, such as the threat of massive tariffs on imported goods in the US, layoffs intended to undo pandemic-era overhiring, and policies that empower employers at …
Would it help to indicate that AI is also perfectly usable for replacing decision makers?
From a shareholder perspective they're the ones actually costing a lot of money, so replacing them with AI seems to provide the best ROI as few AIs have need for yachts, personal planes and paid lunches..
No. AI is a tool for the rich to force the price of the poor down...poor people are about to get a lot poorer.
On flipside though, at least the poor get to be cyborgs first...with their shiny AI powered DeWalt endoskeletons to help them work in construction, their DeLonghi AI powered arms for working in the kitchen and Dyson powered AI lung replacements and legs to help with cleaning the guest wing.
Us techies will be the middle class "pleb mechanics".
Decision makers are also the people making them all the money. More than they are losing them. Much more.
Logic and reason are fine for optimizing established offerings, but they’re anathema for growth. The opportunism that creates huge profits is very often based on factors that information systems have no exposure to and wouldn’t know what to do with if they did.
It’s the same basic principle behind the need for in person meetings instead of just videoconferences. You can do basic, tactical things via video, but the huge coups that create huge wealth require in person meetings that allow participants to negotiate using intangible information that only conveys in person.
Middle management and soft technical specialists will be wiped out by AI. The primary purpose of middle management is as an ablative shield for senior management. AI will become the new shield. Soft technical specialists spend most of their time fixing problems building on top of other systems. Something AI will be much faster at. And let’s face it, software developers set themselves up for replacement by going down the road of constant updates and lowering expectations of users while simultaneously expecting more money. Nothing can save them.
But the companies will still need the decision makers to take the risks based on illogical, wildly irrational, and entirely subjective predictions. That’s where the money is and that won’t change.
"Decision makers are also the people making them all the money. More than they are losing them. Much more."
The role of a CEO is to ensure that the spoils of a profitable enterprise ends up in the pockets of the owners/shareholders.
Without a focused psychopath at the helm, some of the profits might end up in the pockets of the employees or could be used to improve working and life conditions of the employees.
That function of a CEO is priceless.
Every time technology leaps forward, there's always a chorus predicting disaster - usually sung loudest by those whose main skill is easily replaced. Remember when cars were supposed to leave every stable worker and carriage driver jobless? Or when the Luddites insisted machinery would permanently wipe out employment? Yet each time, society adapted, new roles emerged, and innovation prevailed. But CEO of AI company losing his job to LLM would be a peak irony.
Indeed, in future I am going to type out my comments on paper and mail them into the Register with a stamp. Give jobs to postal workers .
Actually, I think parchment is a better option, support farming.
Or should I support masons and have the comment carved into a stone tablet?
I suggest oral transmission1.
The ancient Hellenic bards and the Druids, among others, had the right idea. They believed that transmission of wisdom by writing was prone to misinterpretation and I'm not inclined to argue with someone who could recite 15,693 lines of dactylic hexameter from memory, considering I have a hard time remembering my own cell phone number.
Or do what the Incas did and use quipus knotted cords and string.
_______________
1 I mean with spoken words. Get your mind out of the gutter.
I think the problem this time is that the jobs that will be replaced will be the bullshit jobs (Bullshit Jobs - David Graeber) and the people doing them aren't suitable to do other things.
On the upside maybe we're heading for a post scarcity economy. I want my StarTrek make things out of thin air machines!
Hardly surprising in the UK case. When towns grow up and are situated solely because of one industry, it's no surprise that they struggle when that industry is replaced. Why do we even bother to keep them? Detroit is planning to shrink, and we should have ploughed up most mining villages and the whole of Corby years ago.
The other guy has a valid point. LLMs are not machinery. They are replacing the people who built and operate the machines.
Consumer technology has lowered the expectations of society so far that people are going to be fine accepting the crap experience LLMs deliver. Combined with the hyper selfishness engendered by popular right wing ideologies that have turned individuals into enemies and deified large corporations and LLMs are the vanguard of a nightmare scenario.
Historically, the technological advances you’re talking about were compensated for through direct government intervention. Cost controls, preposterous defense spending, private infrastructure grants, government funded research, and social programs have propped up western economies from WWI onward. Now those programs are being disposed of in favor of companies with monopolistic mentalities that are also the same companies who are pushing LLMs into everything.
It’s a bad situation. I feel sorry for the tech community. They’re going to be vaporized as will healthcare and customer service roles. All the while costs will continue to rise. The only upside is that the people responsible for this disaster will be the first ones to lose their jobs.
"It’s a bad situation. I feel sorry for the tech community. They’re going to be vaporized."
There are two sides to this coin. There are those that train to enter the industry with a specialism and try to dine on it forever, and there are those that constantly adapt to changing tech...the former, are fucked. The latter, will just adapt again as they have done many times before.
>Or when the Luddites insisted machinery would permanently wipe out employment?
No, I don't remember that. Maybe because I studied actual history, not what gets continually repeated as it provides suitable propaganda for a certain way of looking at society. The Luddites insisted no such thing. They insisted that machinery was being used to undermine their control over their own time and labour through centralisation of production and power and that it would further extend the already huge gap in the balance of power between them and the textile merchants become factory owners. They insisted that the uses to which the new technology was being put would not be in their best interests but in the interests of a small elite who would have ever more control and ever more of the profits of their labour. They were, go figure, quite right. Where does this naive idea that technology springs out of thin air come from? What directions technologies take (even what basic research is prioritised), how they are used, what is pursued and what dropped, how implementations are chosen, all these decisions take place within the social, economic and political power relations of a given, existing society. Alternatives to the centralised factory - which handed all power to capital and reduced labour to a mere commodity - were proposed at the time. That they were squashed by the rich elite of merchants become the only owners of the means of production is not a surprise and not something which is directly derived from the technology at all.
It is intended to make CEOs everywhere else worry "am I being stupid by hiring people instead of investing every penny I can into implementing AI across my company?"
I don't know what the "false fret" naming equivalent of a humblebrag is, but that's what this is.
I agree - recently I needed to replace the little solar powered water feature (running over the lead with a lawnmower is not recommended) and thought I'd try the AI search feature. Lots of "try a garden centre" type suggestions but nothing actually useful.
However, AI/LLMs would be ideal for the large number of bullshit jobs out there.
One of my tasks is writing my employer's annual report, something you'd certainly put into the category of a bullshit job. And you know how much use AI is? Sweet FA. I've used both full-fat Copilot, and Claude, and all these seem able to do is parse and summarise existing documents, or answer trivial web-based queries with zero guarantees of accurately. A particularly high risk situation is where the model attempts to combine the shit it has been trained on with internal data, and that's when complete lies is a high likelihood.
AI remains like FSD cars: Pointless until it can be trusted.
To be fair, my colleagues did use ML to screen many tens of thousands of documents to see which tiny, tiny number fit certain complex criteria (circa 0.01%), a job that we didn't have the resource to do manually. It did appear to work, but the unknown is whether there were missed items in the 50k plus documents that AI decided were not relevant.
<.....".....all these seem able to do is parse and summarise existing documents....."....>
That is what I would expect - that is after all, what they are basically designed to do.
It really <isn't> Artificial Intelligence, in spite of the fact that ever more people seem to be being conned into believing that it is.
FSD cars can be trusted...there are numerous videos of it being demonstrated on Youtube on real roads, in real traffic etc etc. The only thing that makes FSD dangerous is the humans on the road doing dumb unpredictable shit, like opening their door into traffic, not following the rules, randomly changing lanes, jack knifing trucks etc.
What I find insane about people and self driving cars is that people are more than happy to get on a rollercoaster, often decades old and subject to forces beyond their comprehension, in a theme park that puts their entire life in the hands of a teenager checking the safety bar and controlling the stop/start button which activates essentially a guided missile...but they won't get in a self driving car that for most of the journey will be going at less than 30mph, which has air bags, seat belts and a chassis designed for high speed collisions.
Also, people get kidnapped, raped, killed and scammed in taxis the world over, all the time...it's probably much riskier to get in a taxi driven by a human than a car controlled by AI and a bunch of sensors.
FSD cars can be trusted...there are numerous videos of it being demonstrated on Youtube on real roads, in real traffic etc etc. The only thing that makes FSD dangerous is the humans on the road doing dumb unpredictable shit, like opening their door into traffic, not following the rules, randomly changing lanes, jack knifing trucks etc.
FSD is not now and will likely not for quite some time be "trusted", not because it cannot handle day to day normal events, but rather because it cannot handle humans on the road doing dumb unpredictable shit. That, sir, is the POINT, humans do dumb unpredictable shit.
You've oversold the state of FSD, unless you're referring specifically to Tesla's FSD (we should probably use a different term for anyone else's as FSD is specifically Tesla's name for it and they're lying). If you are including Tesla's, you're just wrong.
The reason it isn't trusted is the same reason people do trust amusement park rides. Random stuff isn't supposed to happen in front of the rides, and the tracks are designed to make sure they follow the path set for them, which has been tested, and that nothing suddenly appears in front of them. On normal roads, you don't have that certainty. Therefore, if you're going to operate a machine on those roads, you want one that can handle it when cars do things that are unexpected. If we had a parallel network of roads where no humans were operating, maybe more automation could work. We don't have one, which means that you have three choices:
1. Automatic driving has to handle cars driven by humans. People in the automated cars prefer living to dying, but the car has someone else to blame for their death. If that's what's happening, people won't use the cars.
2. Automatic driving isn't allowed because nobody's figured out the liability involved in machines that can't handle the randomness of road travel being placed in the midst of the randomness of road travel.
3. You need to convince somebody to build a parallel road network. Good luck.
I support option 1, and I think we can get there. We will never get there if you insist that it already works. People will see that it doesn't, and if you're not careful, fearful people will force you into option 2. Smart people are working hard to try to accomplish option 1. Don't let them down with arguments like this.
"Automatic driving isn't allowed because nobody's figured out the liability involved in machines that can't handle the randomness"
That's assuming that people own cars in the future. I don't think they do. In which case the liability lies with the manufacturer / operator...
In the future, all cars made by Toyota will be owned by Toyota for example. You will hold a pass granting you accessing their local fleet that either you pay for, or the government pays for (depending on your circumstances).
I don't think in the future anyone will own a car, car manufacturers will not be producing cars to be sold. They will be producing cars that always operate that will be used ad hoc. How that is paid for will likely vary from country to country. In places like Scandinavia, it will be government funded. In the UK it will likely replace bus services. In the US, you'll pay per minute and have different tiers depending on the seat upholstry, whether you use the aircon etc etc etc.
Not owning a car and being able to access whatever is nearby will increase peoples wealth because it will no longer be necessary to own a very expensive liability that you need to insure etc. Driving licenses will be centred around emergency procedures to shut down a vehicle safely in the case of an emergency and what to do if there is a technical fault in order to make the vehicle "safe".
I think this would be awesome, because I wouldn't have to have a driveway anymore...I could turn my drive into a nice green space. I'd be about £8k a year better off (at least).
And why do you believe any of that? Because it would be easy for car manufacturers to do that now: not sell any of their cars, but merely lease them, or in fact operate a system where, when you need a car, you go pick up the nearest one with an electronic unlocking system, drive it to your destination, and leave it. They're not doing that. They're not doing that for a reason: people are willing to pay more to own a car than they are to temporarily use one. If the cars drive themselves, the benefit of owning one is not reduced, and many of the biggest problems are not solved. You might be able to get one if your driveway is located in a sufficiently dense city, but people who don't might value having one near their house, guaranteed, rather than hoping that one will eventually arrive. While it makes the system of operating them as a fleet a little easier, and I'm sure someone would try it, there's no reason to expect they would try to end the concept of ownership, especially when the revenue from repairing old ones is so nice.
Also, what does that have to do with any of this? The problem is with safety or user perception of safety, and the specific part you replied to was about what happens if people become concerned that improper development has caused automatic driving systems to operate unsafely too near to their bodies for comfort. What happens is they get angry and ban them, and then they don't operate, whether owned, leased, or fleeted out. So whether you want automatic driving or you plan to eliminate your driveway, you want to be careful to not sell incomplete automatic driving as more functional than it is.
It is simultaneously "nearly magical in its world-changing abilities" whilst it's "not made a significant difference in employment or earnings in any occupation".
Companies wont be happy when they find out they've spent billions, will have to spend billions more and can't fire whole swaths of their expensive meatsacks.
Unless of course all this is marketing BS...
So, AI is basically The Devil like in many folk stories, in which the devil first poses a bad dilemma to the protagonists, but is then defeated by some cunning plan.
E.g., the devil helps the people to build a bridge across a river,
but demands the first x souls that cross the bridge.
A cunning animal herder, then drives his herd across the bridge,
as it opened.
Local folklore is full of stories like these, and I see some similarity.
One the one hand, billions are invested and the world is ruined in the process (AI doesn't power itself),
because of all the world-changing wonders, respectively, the existential threats if we don't master AI.
But then, in the end, all we get is a Shitty Nazi Clippy that hallucinates and tells lies.
The devil got the herder's soul as the devil and the contemporary theologians would have agreed cattle are not possessed of souls. Pick the right period of history you could have driven the village's wenches over the bridge or some ethnic minority.
Just recently some theologians when asked whether Neanderthals or Denisovans possessed souls they came to the insane conclusion they didn't possess rational souls but potentially those of higher animals.
Given the quantity of Neanderthal and Denisovan DNA spread around our genome we might appreciate our distant ancestors were rather partial to a bit of bestiality on the side. ;)
Is this even up for debate? MS recently did a huge layoff just to pay for the AI that will be used to layoff the rest of the meatbags.
Boards are salivating at the idea of a 'cheap' AI subscription paying off whole rooms of humans and businesses are lining up to sell that dream. No one is saying out loud, "this agentic AI wil replace all your proles" but that is clearly the selling pitch.
This report seems to imply that 'opportunities' for those displaced will be created, but are these like-for-like? Or is the 'opportunity' the chance for your mid-level developer to retrain as a street sweeper or labourer.
TBH I see this happening no matter what. The Board has seen the savings (and bonuses) dangled in front of them, it doesn't matter if the AI only works 25% of the time, or 1 time in 1000 it issues a random kill order on a patient, they're going to buy it anyway, they're going to sack as many as possible anyway, they're going to take the money and not give a shit that everything burns in 2Q.
The only thing that I can see slowing them is that they won't be able to issue a RTO mandate to the AI so the office will be empty. I suspect they'll get over it and push for the return of the workhouse which they can change their now emply offices into. The gov will then pay them to house the ex-workers they paid off who will make batteries for the Tesla Optimus that will replace them in every role they were supposed to retrain for. Win win.
This.
The LLM companies have failed to create a single "must have" product that uses their technology, and they are losing money at a mind boggling rate with no end in sight, so their sales focus now is to convince businesses that they can replace large numbers of workers with LLM 'agents' and the like. Whether anyone cares that these agents cannot reliably replace said workers is doubtful; recent history shows their solution to complaints will be to pass laws stating that "AI" software cannot be challenged or required to explain the process by which it generates any specific output.
Certainly that seems to be the way UK Gov. PLC is leaning. They're trying to legalise pirating by AI companies which 'The Lords' are currently blocking.
They'll win eventually because of the Parliament act after which I plan on incorporating an "ACME AI Ltd" sole trader just for the purposes of 'scanning' copyright content.
Totes legit.
This statement "Workers will need to learn new skills as AI tools spread, but that's always been the case when technology reshapes society." is the key one.
Assume for a moment that AI will change the economy and reduce certain categories of work.
There are several countries where such transformations were done (France, Germany, Belgium post mining and car factory shutdown), where the government either makes it mandatory on large layoffs to provide retraining, or does it from govt side.
When you do that, people can find new jobs with relative ease.
But hoping that peope will somehow find the time to 'upskill' between their second and third jobs is naive.
It has not worked in the US rustbelt, and then that scenario is used to caution against changing the economy.
If you have a long term policy (I know, I'm dreaming), and you have a government and employment agencies that give people the time to retrain, then your economy will rapidly shift without people missing out.
If you don't, only those who can spare the time and money to do so will make it.
It's a great way to create a populist dream, a 'panem & circenses' populace that is 100% dependent on favors of whomever seizes power. Not the best of times.
That's because it's a platitude to stop you kicking off and doing anything about it.
"You're being laid off from XYZ but don't worry we're going to put in huge investment in the area into attracting new businesses, with transport links, training, blackjack and hookers".
A year later
"Due to the economic conditions we're unable to provide any funding, transport or training, but we are totally committed to the area and will demand that industry XYZ contribute"
2 years later
"I'm offksi but having taking up a directorship with XYZ I can confirm they really are very very concerned about the area even though they haven't coughed a bent penny to help, are moving everything offshore and have grifted an extra £2M from the taxpayer for nothing"
A year later
"Vote for us it's all their fault and we reealy reealy promise to make it better", "No vote for us, yes we took donations from XYZ and let them asset strip the region but we reealy reealy promise to make it better this time"
1 month later
"get a job you oik".
What?! Spend money on the meatsack AND the AI? Not happening
The only way that works is if the AI makes the junior produce the same as the senior. Then I can get rid of the senior expensive guys.
If it makes the expensive guys 50% faster then I can get rid of 1 in 3 of them too.
Another snake oil peddler trying to pump stock prices by pretending AI is anything other than a over hyped predictive text machine.
It's "AI in the future will..."
What can it do know you sack of shit?
Scan email.
Make generic dog shit images that are swapping the web.
Make vomit inducing videos
Oh I know, make narcissistic scumbags richer. Yup that's the one.
In Russia people wont say anything bad about Putin in public because they are scared about falling from a window. In America, people are so brainwashed they dont even dare to ever say anything remotely negative about leadership or ceos. Thats how we got Trump and the global financial crisis, the homless plague and more.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
