Forum | U.S. Productivity - an AI Update by nrb1985

Forum index | Previous Thread | Next thread

U.S. Productivity - an AI Update 08:54 - Jan 9 with 2681 views

nrb1985

U.S. productivity hits highest level in 2 years - tentative signs I'd say that AI is beginning to have a meaningful impact in the work place.

https://www.reuters.com/busine

This chimes with current anomaly in U.S. economic backdrop - where earnings and GDP are surging but are in a very low hire/low fire labour environment.

Appreciate concerns around job displacement (particularly for knowledge workers like me!) but for a country like ours that has low productivity, awful demographics, slowing labour force participation - and has decided it wants less immigration - these types of technological advancements may prove extremely beneficial.

[Post edited 11 Jan 19:41]

U.S. Productivity - an AI Update on 12:10 - Jan 9 with 1779 views

J2BLUE

Agree. Lots of stories on social media of people being laid off and told the reason is AI.

We need a plan for 3-4 years time when it's a massive issue.

	Truly impaired.
Poll:	Will you buying a Super Blues membership?

U.S. Productivity - an AI Update on 12:32 - Jan 9 with 1732 views

RadioOrwell

Who do you think will benefit from the increased productivity generated by AI ?
I'm really not sure be me or anyone I know.

U.S. Productivity - an AI Update on 13:03 - Jan 9 with 1659 views

J2BLUE

U.S. Productivity - an AI Update on 12:32 - Jan 9 by RadioOrwell
Who do you think will benefit from the increased productivity generated by AI ? I'm really not sure be me or anyone I know.

Exactly, which is why we need to plan now. It should have started with self service tills. For every human staff member replaced they should have paid 25% of the savings into a national fund to prepare for this. I've been saying this for years.

	Truly impaired.
Poll:	Will you buying a Super Blues membership?

U.S. Productivity - an AI Update on 13:12 - Jan 9 with 1635 views

nrb1985

U.S. Productivity - an AI Update on 12:32 - Jan 9 by RadioOrwell
Who do you think will benefit from the increased productivity generated by AI ? I'm really not sure be me or anyone I know.

I mean, right off the bat mate…

Higher GDP and a bigger tax take without raising tax rates — which can fund public services or help stabilise / pay down national debt.

For the NHS, that could mean a more efficient, higher-quality service without asking everyone to pay more tax.

It could help preserve the welfare state despite awful demographics. Compared with 20 years ago, there are materially fewer working people supporting each retiree — and that ratio is only heading one way.

It also makes lower immigration more economically viable, if that’s the political choice — easing pressure (allegedly) on housing, infrastructure and services without killing growth.

I’m sure I could think of a few more if I really put my mind to it — but it’s after 1pm on a Friday...

U.S. Productivity - an AI Update on 13:17 - Jan 9 with 1617 views

SuperKieranMcKenna

U.S. Productivity - an AI Update on 13:12 - Jan 9 by nrb1985

It would only be a benefit if it’s per capita gdp growth which hasn’t been the case for most of the last decade. Gross output is an irrelevant metric (I.e we have a bigger economy than Switzerland but aren’t wealthier by any metric.

Not convinced tax take will be higher - easier for firms to offshore additional capital than it is for the tax paying workers that are replaced.

U.S. Productivity - an AI Update on 13:20 - Jan 9 with 1604 views

eireblue

U.S. Productivity - an AI Update on 13:12 - Jan 9 by nrb1985

Productivity can simply mean, same output, reduced input.

An AI bot, doesn’t pay income tax or NI.

So if a company generates the same profit with less staff, that doesn’t help the tax income for the Government, it reduces it.

U.S. Productivity - an AI Update on 13:24 - Jan 9 with 1591 views

DarkBrandon

U.S. Productivity - an AI Update on 12:10 - Jan 9 by J2BLUE
Agree. Lots of stories on social media of people being laid off and told the reason is AI. We need a plan for 3-4 years time when it's a massive issue.

It is a pretty big issue now, especially for people graduating at the moment.

I agree a plan would be great ... but I've not heard of any ideas, nor can I think of any.

U.S. Productivity - an AI Update on 13:27 - Jan 9 with 1575 views

DarkBrandon

U.S. Productivity - an AI Update on 13:03 - Jan 9 by J2BLUE
Exactly, which is why we need to plan now. It should have started with self service tills. For every human staff member replaced they should have paid 25% of the savings into a national fund to prepare for this. I've been saying this for years.

25% of the savings isn't much.

If you lay off 8 people, and get self service tills which cost half as much, the 25% will get the state the salary of one staff member.

But that will be more than lost in the increased benefits bill to cover the 8 people.

What we need isn't (just) money to cover the expenses of the people laid off. It is jobs for them to do.

U.S. Productivity - an AI Update on 14:00 - Jan 9 with 1508 views

NedPlimpton

U.S. Productivity - an AI Update on 13:03 - Jan 9 by J2BLUE
Exactly, which is why we need to plan now. It should have started with self service tills. For every human staff member replaced they should have paid 25% of the savings into a national fund to prepare for this. I've been saying this for years.

How on earth do you calculate the saving from a self serve checkout?

For a start there's higher shrinkage from self-serve so the lost stock would be a factor, plus there's the infrastructure, maintenance, software costs behind it. There's also the likelihood that some price increases in food are offset by savings in staff costs etc

It's a bit simplistic to think that x number of self serve checkouts means x number of people are without a job. Same with loads of other systems and tech based initiatives. They may replace some frontline workers, but the the software industries themselves employ people. Do you count the people who will be in a job because of the contract their company has providing Tesco with tills in the equation?

[Post edited 9 Jan 14:12]

U.S. Productivity - an AI Update on 14:16 - Jan 9 with 1458 views

RadioOrwell

U.S. Productivity - an AI Update on 13:12 - Jan 9 by nrb1985

1. Higher GDP does in no way mean bigger tax take considering how the rich love avoiding tax.
2. You're dreaming.
3. See 2.
4. Immigrants are the ones doing the jobs AI can't do.

U.S. Productivity - an AI Update on 14:38 - Jan 9 with 1411 views

nrb1985

U.S. Productivity - an AI Update on 14:16 - Jan 9 by RadioOrwell
1. Higher GDP does in no way mean bigger tax take considering how the rich love avoiding tax. 2. You're dreaming. 3. See 2. 4. Immigrants are the ones doing the jobs AI can't do.

A very detailed and thoughtful repost there...

I can see you've put a lot of thought into things.

U.S. Productivity - an AI Update on 16:29 - Jan 9 with 1290 views

reusersfreekicks

U.S. Productivity - an AI Update on 13:12 - Jan 9 by nrb1985

And back on planet earth....

U.S. Productivity - an AI Update on 16:48 - Jan 9 with 1262 views

Ewan_Oozami

U.S. Productivity - an AI Update on 13:12 - Jan 9 by nrb1985

Are you Peter Thiel in disguise?

If GDP is increased because govts are spending more taxpayer money on AI-related infrastructure instead of, oh I don't know, universal healthcare, is that really a benefit to humanity?

https://gizmodo.com/us-awards-

	You are the obsolete SRN4 to my Fairey Rotodyne....
Poll:	What else could go on top of the cake apart from icing and a cherry?

U.S. Productivity - an AI Update on 17:05 - Jan 9 with 1209 views

nrb1985

U.S. Productivity - an AI Update on 16:48 - Jan 9 by Ewan_Oozami
Are you Peter Thiel in disguise? If GDP is increased because govts are spending more taxpayer money on AI-related infrastructure instead of, oh I don't know, universal healthcare, is that really a benefit to humanity? https://gizmodo.com/us-awards-

Eh?

I’m talking about private-sector AI adoption raising productivity — firms producing more with the same inputs — which lifts GDP and the tax base without higher public spending.

Think you've conflated two different things here.

U.S. Productivity - an AI Update on 17:07 - Jan 9 with 1203 views

nrb1985

U.S. Productivity - an AI Update on 16:29 - Jan 9 by reusersfreekicks
And back on planet earth....

again, another excellent repost.

U.S. Productivity - an AI Update on 17:14 - Jan 9 with 1193 views

RadioOrwell

U.S. Productivity - an AI Update on 17:05 - Jan 9 by nrb1985
Eh? I’m talking about private-sector AI adoption raising productivity — firms producing more with the same inputs — which lifts GDP and the tax base without higher public spending. Think you've conflated two different things here.

Ground control to Major Tom

U.S. Productivity - an AI Update on 17:48 - Jan 9 with 1144 views

CoachRob

U.S. Productivity - an AI Update on 17:05 - Jan 9 by nrb1985
Eh? I’m talking about private-sector AI adoption raising productivity — firms producing more with the same inputs — which lifts GDP and the tax base without higher public spending. Think you've conflated two different things here.

Your definition of productivity is the one used by mainstream economists and it is complete nonsense.

You could search for this using any LLM such as Co-Pilot;

"Here’s the short version: mainstream economics treats productivity as a purely monetary or labour–capital phenomenon, while thermodynamics shows that physical energy flows and entropy fundamentally constrain all production. When you ignore those constraints, you systematically mis-measure what actually drives economic output.

1. Standard economics treats energy as “just another input”
In neoclassical growth models (like Cobb–Douglas), the contribution of a production factor is assumed to equal its cost share. Since energy has historically been cheap—often only 5% of total costs—economists conclude it contributes only 5% to output.

But this is a category error.

Energy is not like labour or capital. It is the physical capacity to do work, and without it, neither labour nor capital can produce anything at all. As Ayres and Kümmel argue, energy conversion in machines is the real basis of industrial growth.

2. Thermodynamics says: no energy, no production
Thermodynamics gives us two non-negotiable laws:

1st law: You can’t produce output without energy inputs.

2nd law: Every transformation increases entropy, meaning useful energy is degraded.

Economists rarely incorporate these laws into productivity analysis. As a result, they attribute output growth to “technology” or “TFP” (total factor productivity), when in reality much of it is simply better energy conversion efficiency.

This is exactly the argument behind thermoeconomics and ecological economics.

3. “Technology” is often just better thermodynamic efficiency
When a steam engine goes from 5% to 20% efficiency, economists call the resulting output jump “technological progress.”
A physicist calls it: more useful work extracted per unit of energy.

By ignoring thermodynamics, economists mislabel physical improvements as abstract “productivity.”

4. Macroeconomics lacks a physical foundation
A growing body of work argues that macroeconomics should be built on thermodynamic principles, not just preferences and markets. For example, Chater & MacKay propose an axiomatic macro theory explicitly modeled on thermodynamics.

This approach yields:

clearer definitions of value and money

better understanding of inflation

more realistic constraints on growth

But mainstream economics still treats the economy as a closed, frictionless system—ignoring entropy, energy flows, and physical limits.

5. Ignoring thermodynamics leads to bad policy
When you treat energy as trivial, you get:

underestimation of the economic impact of energy shocks

overestimation of the role of capital and labor

unrealistic expectations for “green growth”

failure to see that economic growth is ultimately energy growth

This is why ecological economists like Georgescu‑Roegen and Daly argue that ignoring thermodynamics makes mainstream economics fundamentally incomplete."

Why are some people so gullible with these weird definitions, we all did thermodynamics at school, some of us still use it in our own work, but there are these people who think the world runs on magic.

U.S. Productivity - an AI Update on 18:01 - Jan 9 with 1090 views

Ewan_Oozami

U.S. Productivity - an AI Update on 17:05 - Jan 9 by nrb1985
Eh? I’m talking about private-sector AI adoption raising productivity — firms producing more with the same inputs — which lifts GDP and the tax base without higher public spending. Think you've conflated two different things here.

So sorry, I based my response on an AI summary of your post...

	You are the obsolete SRN4 to my Fairey Rotodyne....
Poll:	What else could go on top of the cake apart from icing and a cherry?

U.S. Productivity - an AI Update on 18:06 - Jan 9 with 1079 views

RadioOrwell

U.S. Productivity - an AI Update on 14:38 - Jan 9 by nrb1985
A very detailed and thoughtful repost there... I can see you've put a lot of thought into things.

This is to your 1st point.

That claim sounds very attractive, but it rests on assumptions that often don’t hold in practice. The main problem is that higher GDP from AI does not automatically translate into a reliable or fair tax base. Here’s why.
1. GDP can rise while tax revenue doesn’t
AI-driven growth can be decoupled from taxable activity:
Capital-heavy, labor-light growth: AI boosts profits more than wages. Since profits are often taxed more lightly (and more easily avoided) than labor income, the tax take may rise far less than GDP.
Intangible assets: Much AI value comes from software, IP, and data, which are easier to shift across borders for tax purposes.
Global firms, local costs: AI firms may generate GDP domestically while booking profits abroad.
Result: GDP goes up, but governments don’t capture much of it.
2. The tax burden may shift downward, not upward
Even if total revenue rises, who pays matters:
Fewer workers, weaker payroll taxes: Automation shrinks employment and wage growth, eroding income tax and social security contributions.
Regressive substitution: Governments may rely more on VAT, consumption taxes, or user fees, which hit lower-income households harder.
“Hidden” tax increases: Public services may still be funded, but through mechanisms that feel like higher taxes to ordinary people.
Result: The tax system becomes less progressive even without rate hikes.
3. AI increases inequality, which weakens fiscal stability
High inequality creates fiscal problems:
Political resistance: Concentrated wealth is harder to tax due to lobbying and political power.
Lower aggregate demand: If AI gains accrue to a small elite, consumption grows slowly, limiting VAT and sales tax revenue.
Higher social spending needs: Unemployment support, retraining, healthcare, and social stability costs rise—offsetting any revenue gains.
Result: Net fiscal position may not improve at all.
4. GDP is a poor measure of public benefit in AI economies
AI can inflate GDP without improving public finances:
Cost-cutting vs. value creation: Replacing workers with AI raises productivity and GDP but can reduce tax revenue.
Free or low-price digital services: These increase consumer surplus (counted weakly in GDP) but generate little taxable revenue.
Quality vs. quantity: AI can increase output without improving living standards in ways that reduce public costs.
Result: Governments can look “richer” on paper but struggle to fund services.
5. Timing mismatch: revenue later, costs now
AI creates immediate fiscal pressures:
Worker displacement
Retraining programs
Social safety nets
Regulatory oversight
But tax benefits (if any) arrive later and uncertainly.
Result: Short- to medium-term budget stress, even with long-term growth.
6. Debt dynamics may worsen, not improve
If governments expect AI to “grow them out of debt”:
They may delay hard fiscal choices
Borrow against uncertain future revenue
Underestimate structural changes to the tax base
If AI growth underdelivers fiscally, debt ratios can rise.
7. It assumes strong governance that may not exist
The claim only holds if governments:
Can effectively tax AI-driven profits
Can coordinate internationally on tax rules
Can prevent regulatory capture
Can redistribute gains politically
These are political constraints, not technical ones.
Bottom line
The problem with the “higher GDP, higher tax take without raising rates” argument is that it treats economic growth as fiscally neutral and socially automatic.
In reality, AI growth tends to:
Favor capital over labor
Concentrate income and profits
Exploit tax arbitrage
Increase public spending needs
Without deliberate tax reform and redistribution, AI can raise GDP while leaving governments poorer, public services underfunded, and debt harder—not easier—to manage.

U.S. Productivity - an AI Update on 18:12 - Jan 9 with 1063 views

RadioOrwell

U.S. Productivity - an AI Update on 14:38 - Jan 9 by nrb1985
A very detailed and thoughtful repost there... I can see you've put a lot of thought into things.

To your second point ...

That argument sounds plausible on the surface, but it quietly assumes away several hard constraints. The core problem is that AI-driven productivity does not map cleanly onto labour-intensive public services like the NHS, nor does it fully solve demographic pressure. Here’s where the logic breaks down.
1. NHS productivity is fundamentally constrained by human labour
Most NHS costs are people, not processes:
Care, diagnosis, surgery, nursing, and patient interaction are labour-intensive and often cannot be automated without quality or safety trade-offs.
AI may improve supporting tasks (triage, admin, imaging, scheduling), but these are not where most spending is.
Even if AI makes doctors more productive, you still need the doctor, and demand rises to fill any capacity gains.
Result: Productivity gains are real but bounded, unlike in manufacturing or software.
2. AI often increases demand, not reduces cost
This is a classic “efficiency paradox” in healthcare:
Faster diagnosis → more people diagnosed
Better screening → more treatment pathways
Improved survival → more long-term care
Lower per-unit cost → higher total usage
Healthcare demand is elastic upward.
Result: Quality improves, but total spending often rises, not falls.
3. Demographics hurt the NHS twice
Ageing populations don’t just reduce the worker-to-retiree ratio:
Fewer taxpayers relative to patients
Much higher per-capita healthcare costs for the elderly
AI may:
Help manage chronic disease
Improve efficiency at the margin
But it does not reverse:
Multi-morbid ageing
End-of-life care costs
Long-term social care needs (mostly non-medical, non-automatable)
Result: Demographics remain a dominant cost driver.
4. Public sector productivity ≠ private sector productivity
The argument borrows intuition from the private sector, but the NHS faces constraints that firms don’t:
No profit motive: Savings don’t automatically reinvest or reduce tax needs.
Political accountability: “Efficiency savings” often get redeployed, not banked.
Baumol’s cost disease: Wages in healthcare rise in line with the broader economy even when productivity doesn’t.
AI does not repeal Baumol’s disease; at best, it slows it slightly.
5. Implementation costs are large and ongoing
AI in the NHS isn’t free productivity:
Procurement and licensing costs
Integration with legacy IT
Cybersecurity risks
Training and workforce adaptation
Regulation, validation, and liability management
These are recurring costs, not one-off investments.
Result: Net savings are smaller than headline productivity claims.
6. AI does not fix the tax base problem
Even if AI raises economy-wide productivity:
NHS funding depends on taxable income
AI growth skews toward capital and high earners
Payroll taxes and income tax weaken as labour share falls
So the fiscal capacity to fund the NHS may still erode.
Result: Better tools, weaker funding base.
7. Political economy problem: expectations outrun reality
Perhaps the most dangerous issue:
Policymakers may assume AI will “fix” NHS finances
Necessary tax, staffing, or structural reforms get delayed
When gains fall short, the system faces sharper crises
AI becomes a substitute for hard choices, not a complement.
Bottom line
AI can absolutely:
Improve NHS quality
Reduce admin burden
Help clinicians make better decisions
Partially offset workforce shortages
But it cannot:
Replace the human core of care
Neutralise ageing demographics
Sustain a tax-funded welfare state on its own
Deliver “more and better healthcare without higher taxes”
The claim is best understood as “AI can buy time and reduce pain”, not “AI preserves the welfare state for free.”

U.S. Productivity - an AI Update on 18:20 - Jan 9 with 1039 views

Ewan_Oozami

U.S. Productivity - an AI Update on 18:12 - Jan 9 by RadioOrwell

That is some top "productivity" there!

	You are the obsolete SRN4 to my Fairey Rotodyne....
Poll:	What else could go on top of the cake apart from icing and a cherry?

U.S. Productivity - an AI Update on 19:38 - Jan 9 with 987 views

J2BLUE

U.S. Productivity - an AI Update on 14:00 - Jan 9 by NedPlimpton

[Post edited 9 Jan 14:12]

It's a starting position idea not a government bill. I set it low at 25% for that reason. Even if it was 10% it would be some income for the government. In the fantasy world we all share in the benefits of new technology and AI, especially when it will cost millions of jobs. Or we could do nothing, concentrate wealth in even fewer hands and have the majority of the population living in poverty. We're going to have to fight for a share of the benefits of new tech and AI.

	Truly impaired.
Poll:	Will you buying a Super Blues membership?

U.S. Productivity - an AI Update on 20:31 - Jan 9 with 951 views

glasso

Fantastic! Can't wait until we're all unemployed and a couple of tech bros own the world.

U.S. Productivity - an AI Update on 21:01 - Jan 9 with 908 views

thebooks

U.S. Productivity - an AI Update on 13:12 - Jan 9 by nrb1985

This is nonsense, pure whatiffery.

US productivity is a pyramid scheme of everyone building vast data centres and shovelling huge amounts of money to Nvidia.

U.S. Productivity - an AI Update on 21:08 - Jan 9 with 901 views

NedPlimpton

U.S. Productivity - an AI Update on 19:38 - Jan 9 by J2BLUE

But that's the point. You can't just say for every job that software has replaced X amount must go to the government. It's basically a tax on innovation and where do you draw the line? What about e-tickets and scanners at turnstiles instead of people? What about emails instead of letters delivered by postmen?

Our main sponsor is a workflow automation company. What happens to their employees when the cost of providing their services goes up because their customers have to give 10% to the government?

Forum index | Previous Thread | Next thread