No wait...AI productivity is just around the corner....we just need $ trillions more investment, more power, more memory, more data centers...you'll see...we'll replace that $20 an hour wage white collar slave...

> improved by 2.7 per cent annually from 1947 to 1973, but just 2.1 percent between 1990 and 2001

Was going to make a pointed remark about productivity in the computer programming sector improving after, you know, we all actually got hold of enough computers to program.

The it became obvious - as more programmers became able to, well, *be* programmers from the 1970s onwards[1] this overlapped with our being able to play increasingly good video games. And we couldn't even claim the PCs were compiling in the background. So down went individual productivity...

[1] anyone tries to claim "PCs didn't exist before the IBM PC will get viciously sneered at

The productivity gain was small in monetary terms because the things being produced got a lot cheaper.

They got cheaper because you cannot sell things to the masses if they are too expensive for the masses to actually buy.

This is also what the AI bubble has entirely forgotten.

To add more complexity, economic growth and productivity figures measure things in cost/price, not capability.

So the transition from crappy old CRT monitors to flat panels does capture the fact that on an inflation adjusted basis the cost has come down, so lower output, and it doesn't capture the value to me that my display is much bigger, MUCH better quality even than my one time Iiyama CRT, takes up less space, uses a fraction of the energy. In the case of video games power use and hardware needs have gone up, but contrast the original Wolfenstein running off floppies to any modern FPS, and there is not really a comparison is there? Yet still the cost per game has mostly hovered in the region £30-40 (ignoring modern free to play + in game purchases). Same technology driven progress is true for household lighting, vacuum cleaners, and a million and one other things that we take for granted - cordless anythings, the variety, price and quality of streaming services. Contrast the high price and iffy quality of Sky or Virgin multi-channel packages of yore with any competent modern day streaming service. Consider the benefits of both mobile connectivity and the internet. These changes have affected almost every aspect of how we live. Even medicine can now keep people alive until their brain rots.

So if growth and productivity measures allowed for capability, and the fact that many (not all) goods and services costs have fallen in real terms, over the past fifty years we've been the beneficiaries of the most remarkable period of innovation and growth since the early days of the industrial revolution. At the root of that have indeed been computers - spreadsheets to support fast, accurate analysis in fields from manufacturing to medicine, computer models to optimise efficiency, automation, and either supporting or creating markets for new innovations from lighting to magnets to pharmaceuticals. Whilst the West may have regulated itself into a cul-de-sac, China can build a new nuclear power plant from scratch in just over four years - some of that's labour control and permitting, but most of that speed is down to technology enabling things to be designed, planned, delivered and built quickly.

Whether AI will continue that trend I can't say, and I'm currently sceptical, but to assert that there's been low or declining growth or productivity over the past half century depends entirely on how it is measured. I acknowledge the down sides (social media, Musk, misuse of AI tools, creation of the Deliveroo under-class) but in net terms so much has been achieved that we don't measure and take for granted.

Quick poll, for those of us who can remember what life was like in 1974, hands up those who (at their current age) would wish the world to flip back to that? Not me, I can assure you.

Whether AI will continue that trend I can't say

Narrator: AI did not continue that trend.

I find the best way to use AI is to get it to provide me with a list of useful reference articles, which I can then read and (hopefully) understand and then go use the information constructively. If I ask it to generate code then I need to spend time checking that code, if I want it to write prose then I have to spend time proofreading it and checking for accuracy. I've seen it be flat wrong before now. One day it may work well enough to just get on and do stuff, but I remain to be convinced that it is reliable outside of fairly narrow areas where it's been trained on carefully-curated content.

Even the list of reference articles is likely to include some that are completely made up; there is nothing to connect the text generated by a text generator to the real world even if everything it was trained on was accurate - it's still building it's own statistical connections, and that's going to lead to unpredictable results.

LLMs are not designed to produce a CORRECT result. They are designed to produce A result. If a query exhausts a line of “results”, it will default to the …well …default. Which could be anything

From the linked-to article about the "Solow Paradox"

> Vandenbroucke finds that the share of 23-33 year olds in the economy is strongly negatively correlated with labor productivity growth.

(and then that article continues by showing Vandenbroucke to be optimistic)

So don't try to push your worker age down.

Unless you are trying to show bottom line numbers that go up because you are selling what the recently-fired oldies got through to production-ready status, so now you can try saving on wages with a young workforce doing the design-free bulk production steps (i.e. sit on the phone nagging the factory who are doing the injection moulding). Of course, nobody would be stupid enough to do that, to try and game actual productivity against wages, as it couldn't possibly be sustainable.

Not in the least bit sustainable, so once you've published those figures and the company has been bought up on that basis, you - won't give a damn, aboard the new yacht.

PS

What? You wanted some comment about AI?

Well, as the article shows, that is your excuse for shedding anyone over 22. What more needs to be said? Except, do it quick whilst people still believe this AI nonsense, otherwise the buyer might catch on.

Yeah, they also say (further down):

> labor productivity growth is driven by [workers] aged 35-54

So the plan should be that we would be granted full retirement and associated benefits by age 55 imho.

Somebody please inform the French government (and yours) of that reproducible scientific fact as [1]they want to force us all to strenuously labor 'til we're at the doorstep of certain death (and taxes), barely able to ambulate without an EMS and staircase lift-chair, thoroughly counterproductive in the daily toil of our routine torture, aged 10-years beyond, at 64.

The optimal course is clearly: 1) enjoy life for 34 years; 2) work productively for 20 years, and; 3) enjoy life again for 34 more years. Anything else is a bloody human trafficking bondage ripoff imho! ;)

[1] https://en.wikipedia.org/wiki/2023_French_pension_reform_law

That's due to the demographic bomb.

Everything done by a government is paid for by the current workers. Pensions, healthcare, defence, education etc, it's all paid for by the people actually working - either at the time, or over the next 10-20 years.

Never by the people who used to work.

This only works if the people working outnumber the people who aren't.

Because families have been getting smaller for a long time, the retired are going to outnumber the working pretty soon unless something drastic happens.

The only ways for rich countries to avoid this are to either accept significant immigration, or to prevent people retiring.

> The only ways for rich countries to avoid this are to either accept significant immigration, or to prevent people retiring.

Ah, if only we had hordes of working-age people who would literally risk death to come and work for us for peanuts... oh, wait.

I am constantly amazed and dismayed by how the West has been shooting itself in the foot on immigration for decades now, and seems to be aiming higher and higher with every election. I wonder if we'll stop before we get to the heart.

If a country's economic model is reliant upon a growing population, then it's a Ponzi scheme really, isn't it? And absent infinite land and other natural resources, it has to break at some point?

In the UK we're reliant upon migrants to staff our health service and do the other hard, dirty, poorly paid jobs that the established population (including the offspring of previous migrants) are reluctant to do. yet we have millions of people who are not working - some claiming state handouts, some not. That doesn't seem like a good argument for migration.

"If a country's economic model is reliant upon a growing population, then it's a Ponzi scheme really, isn't it?"

Of course it is. I don't know about other countries but the UK state pension schemes and Civil Service pension schemes are exactly that.

Sure, but it's not necessarily the case that we need a growing population. It's more the case that, without immigration, the future of all advanced economies involves the *working age* population shrinking rapidly, while the post-working-age population grows for much longer. Nobody has a good answer as to how any country deals with that without substantial hardship and turmoil.

(The bad news is that even immigration can only be a temporary and incomplete 'solution', as fertility rates are dropping like a stone absolutely everywhere; probably within 20-30 years there won't be enough immigrants to "go around" either.)

"the future of all advanced economies involves the *working age* population shrinking rapidly, while the post-working-age population grows for much longer"

Not if the retirement age is raised to 82

> If a country's economic model is reliant upon a growing population, then it's a Ponzi scheme really, isn't it?

The problem isn't the absolute number of people, but the ratio between workers and retired. We don't have to import immigrants because we need more people; we have to import them because we need to raise that ratio. You are correct that a good economic model should be able to work at steady state; unfortunately, we are very much not at steady state, as that ratio is rising.

(^^ last word should be "falling")

Everything done by a government is paid for by the current workers.

So what happens to the income tax I pay on my pension? What happens to the 20% of almost everything I spend which goes to the government as VAT? My pension is deferred wages. Just because I am not working for the money right now does not mean that someone else is working for it any more than someone else is working for you when you're asleep or on holiday.

"1) enjoy life for 34 years; 2) work productively for 20 years"

Not quite that. The 34 years is spent learning to be at peak productivity for the 20 years.

Let's see the productivity gains from 2020 - 2025

And AI doesn't improve productivity.

Some day soon, the Board will have to deal with this issue.

Preferably before the investors sur themselves again (the term "cutting off your nose to spite your face" comes to mind).

PS: on a side note, who was the idiot who decided that investors had the right to sue the company they invested in ? If you don't like what the company is doing, go put your money somewhere else, period.

In theory there are limits to the lies CEOs can tell when they try to attract investors. It would make sense for some kind of penalty for saying you have a working prototype electric truck when all you have is a video of a truck without a power train rolling down a long hill. Claiming you can summon your FSD car from the far end of the car park should also have consequences if the promotional video has lots of prangs edited out.

In practice first round investors like convincing liars. They are useful for attracting replacement investors if the project turns out to be a bust.

"Caveat emptor for thee but not for me"

" who was the idiot who decided that investors had the right to sue the company they invested in ?"

That wasn't the idiot in the case. It would be the lawyers involved and will do quite nicely out of it. The idiots are those who believed them.

I tend to agree on the stock side of things, although when a board is subservient to the ceo (like tesla where the board is elon's lapdog) I can see it. But bonds are a different thing entirely. Oracle just got sued by some bondholders. I don't touch ai bonds. But oracle going back to the slop trough after issuing 18B for AI and then 2 months later issuing another 38B. Well that is securities fraud. Oracle had to know it needed more cash for the AI investment and they flat out lied in the first bond sale by not disclosing it. Ellison should fry for this.

You only have to look at the PC revolution's impact on productivity to have a go at forceasting AI's impact. There's no doubt that that PCs have made some things massively more affordable and improved productivity (e.g. circuit simulation, analysis, requirements management, etc), but they also made life shitter for working engineers. A few examples off the top of my head.

Engineers have taken on the workload that secrataries used to do. In the good old days I could dump my log book on Evie and she'd have a first bash at typing up a test report, give it to me to mark up and then finish it. It would have been a couple of pages of technical content whereas today it will be much thicker and the engineer has to do it all cos there are no more secretaries. Paperwork has become more onerous just because it can; outlining, headers, contents, etc. and all done by engineers who should be working. Ditto meeting minutes, metrics for managers, .....etc.

Powerpoint - nuff said.

Endless Zoom-enabled meetings.

AI will be the same. There will be some focused stuff for which AI will be transformational - lke the protein folding stuff, but for the rest of us it will just shift work from creating and designing things to checking and testing an AI's output then having to argue with the boss about fixing it cos his bonus depends on AI being brilliant and engineers being fired.

I'm allright, Jack, cos I'm old and on my way out, but I feel sorry for today's grads.

I remember that. Another thing my shared sec did was a card with any meetings for the day, arranged meetings. She was also one of the early adopters of FrameMaker, so made document prep so much easier for me. We also had real offices back then. There was something to be said for all those perks.

When I started work as an academic in 1990, I had a one-quarter share of a secretary who did all my admin, correspondence and travel booking for me. By 2010 that was over and for the last fifteen years of my career (sic) I spent a good 50% of my time doing work on a (promoted) academic salary which would have been better, in all senses, done by a clerical/admin worker on a third as much.

"95%" is a nice headline figure, but misinterpreted. This figure in the study specifically is about the transformational impact of specific, bespoke AI projects. The same study also shows significant benefit of generic AI solutions in something like 7 out of 9 industries but I've never seem those one quoted. It's not sensational enough.

We should think twice if the same study is quoted over and over again to support a specific conclusion, rather than being supported by a broad evidence base.

> They’re firing people because of AI, and then three weeks later they hire a team in India because the labour is so much cheaper.

It's Always Indians.

I don't think it's overly cynical to suggest that this is the planned outcome. In most countries you can't just fire your workforce to employ someone cheaper in another country, but if you can find a legitimate reason to downsize your business there's no obligation to rehire the workers you laid off. As an untested technology that nevertheless is having grandiose claims made about it, AI is close to a perfect boondoggle for that. If it works then great, no more paying workers; if it doesn't, then you've got rid of the well paid local workers with rights and can hire a boiler room in Mumbai.

Even without the outsourcing (a better term would be much cheaper workers in places where they dont give a shit what happens to industrial waste)

A lot of rustbelt jobs were doomed when the spree for robotics and CNC machines started.

Why employ 3 machine setters and 12 capstan operators when we can buy one machine to bang out 50 000 parts a year and the only thing we have to do is make sure its fed with material once every couple of hours and swap the program when the current batch is complete. thats what killed the rustbelt.

You have to look at a modern steel rolling mill, maybe 200 people across 3 shifts doing what took 2000 50 yrs ago. all made possible by computers and automation to see this reality.

The trouble with AI is whose job is it going to replace..... unskilled idiots working in factories have already gone... no AI is coming for the middle classes, the managers, the accountants, the HR staff (and us programmers) and they'll find the rewards of AI wont translate into better wages for the survivors... the rewards will all flow upward just as they did when robots began infesting where I work.

Nobody has yet produced Artificial Intelligence stupid enough to take on HR roles. You just can't get silicon to drool.

quote: some of last year’s large-scale job cuts were belt-tightening decisions, not the result of shifting work to AI.

As we mentioned on here. We also pointed out that jobs were cut and outsourced, as the AI software was pants.

One penny that hasn't dropped is that there are ceilings to productivity. You raise them with access to migrant labour, willingness to invest, managerial competence, wealth circulating, a widespread sense of security, and a strong global market, but when these variables are fixed, lowered or declining, all that AI will give you is an additional bill for bollocks software that doesn't work very well.

Step back and re-examine, over time, the cost of using expensive, customised, networked software and doing everything digitally, including endlessly buying new hardware, new software and paying for hacks, against the use of simple, non-networked tech, where required, and paper, not digitising things unless there is real benefit in doing so, not building honeypots of data that are targeted, and not networking all your kit, making it vulnerable. We have developed an expensive addiction to tech, and the only ROI benefit is for the tech sector. We need to reboot how we do things and only use tech that benefits us and our bottom line, where we need it.

It wasn't clear to me from the article what particular measure of productivity was intended. The context suggests labour productivity eg GDP/hours worked/year.

(Productivity is a measure of the output for the required inputs.)

To improve labour productivity either either the output eg GDP must increase or alternatively the inputs eg hours worked must decreased.

The dream was that tech can reduce hours worked but of course tech introduced its own (skilled) labour costs so might be break even.

I would guess the introduction of PCs into the workplace had little effect on hours worked other than they were different hours worked eg typing pools vanished fairly quickly but the (steno)typists morphed into office managers, PAs and a host of other clerical roles—many created by PCs and related technology.

I suspect the 2001-2007 productivity gains were increasing contribution to GDP from the higher value Internet enabled industries and services. Ironically without the advent of ubiquitous personal computing the Internet as we know it would not have been possible (for good or ill.)

The parallel that might be applied to AI/LLM is that total hours worked are likely to remain static but change structurally so that the only prospect for increased productivity is for AI to enable or create high value services or products (which customers want and can use.)

Personally I cannot see AI producing a anything of sufficiently high value to justify the input costs.

A chilling thought is that Grok Nudify might be the zenith of high value AI services which might not be too far from the truth when you consider just how much network traffic is porn and adjacent "adult services" and how big the porn industry is.

... vice president and principal analyst ...

That's about one step up from "makes the tea", isn't it?