TL;DR
A November 12 Thorsten Meyer AI article uses the history of productivity growth to assess AI’s potential effects on work. It cites a study finding a 15% productivity gain for support agents in one setting, while executives estimate AI has added about 0.2 percentage points a year to productivity growth so far. The broader effect on jobs, wages and economy-wide output remains uncertain.
Thorsten Meyer AI published an installment on November 12 examining how artificial intelligence could affect work through productivity, or the value produced per hour worked. The article says reported gains inside individual firms have not yet translated into a large measured contribution to overall productivity growth: it cites an estimate of about 0.2 percentage points a year so far, alongside a study finding a 15% gain for support agents in one setting.
Get comfort and recovery gear delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
The installment, part of the series “Work After Work,” uses the history of earlier technologies to explain why AI’s effect on work may take time to appear in economy-wide statistics. It recalls economist Robert Solow’s 1957 analysis of U.S. output per hour from 1909 to 1949. Solow attributed 87.5% of the increase to technical change and 12.5% to increased use of capital, such as machinery. The article notes that the “technical change” category includes more than inventions; methods, organisation and skills can also affect output.
It also points to the computer era as a precedent for a lag between adoption and measured gains. The article says productivity growth slowed after 1973 even as computers spread, then rose to about 3% a year from 1995 to 2004. It attributes to Federal Reserve economists the estimate that information technology accounted for about two thirds of that acceleration. These historical figures provide context, but they do not establish that AI will follow the same path.
For a current firm-level example, the article cites research reporting a 15% productivity gain among support agents in one study. It also reports that executives put AI’s contribution to productivity growth at about 0.2 percentage points a year so far. The source excerpt does not give the study’s full details or explain the executive estimate’s methodology, so the figures should be read with those limits in mind.
How AI Can Change the Way We Work
Early workplace gains are reported, but AI’s effect on economy-wide productivity, jobs and wages remains uncertain.
What productivity measures
It measures output per hour worked, at a workplace, sector or economy-wide level.
AI may help people complete some work faster.
Adoption and new ways of working can take time to spread.
Economy-wide productivity data show whether gains add up.
Why history offers context
Past productivity shifts show that adoption and measured gains can unfold over time.
What the evidence can tell us
The reported figures describe different settings and do not establish AI’s overall effects.
- A 15% gain for support agents in one study.
- Executives estimate about 0.2 percentage points a year in productivity growth.
- Effects on employment, wages and job quality.
- Whether gains persist across occupations and industries.
What to watch next
Evidence should connect AI adoption to broad workplace outcomes.
- Measured output per hour as firms adopt and reorganise around AI.
- Whether reported gains hold across occupations and industries.
- How saved time, costs and productivity gains affect workers.
Why Economy-Wide Gains Matter
Productivity affects how much an economy can produce from each hour of work. The article argues that sustained gains can support higher living standards over time, while growth that depends only on longer hours or bringing more people into employment has limits. It illustrates the compounding effect with two rates: at 2% annual growth, output per hour doubles in about 35 years; at 1%, it takes about 70 years.
That long horizon matters for workers because new tools can change tasks before they change the total amount produced across an economy. A company may complete particular work faster, yet the gain could be difficult to see in national statistics if adoption is uneven, if other costs rise, or if saved time is used for different tasks. The source does not establish which of these patterns is occurring across workplaces.
AI’s effects also matter beyond efficiency figures. Changes to work could affect which tasks employers assign to people, what skills they seek and how the value created is distributed. The source offers historical productivity measures and a limited firm-level example, but it does not report evidence on AI’s net effects on employment, pay or job quality. Those questions remain separate from whether output per hour rises.
Popular right now

Bose QuietComfort True Wireless Earbuds with Active Noise Cancellation, Bluetooth 5.3, USB-C Charging, IPX4, Black

Ninja Foodi Smart 6-in-1 Indoor Grill and Air Fryer, 4 qt, FG551

Shark Rocket Pet Plus Corded Stick Vacuum, Bordeaux/Silver, HV322
As an affiliate, we earn on qualifying purchases.
Lessons From Earlier Technologies
The article defines labour productivity as output per hour worked. It distinguishes this measure from total factor productivity, an estimate of output growth that cannot be explained by changes in hours worked and capital. Because total factor productivity is calculated as a residual, the source says economists continue to debate what it includes.
Its agricultural example shows how productivity can change the distribution of work across sectors. The source cites U.S. Department of Agriculture data showing farm output in 2023 was 210% higher than in 1948, while combined inputs rose at an average annual rate of 0.12%, or about 9% across the period. It gives average annual total factor productivity growth of 1.40%. These are measures of U.S. farming over that time span, not estimates of AI’s effects.
The article connects those changes with a falling share of workers employed in farming. It reports that the share declined from about one in eight U.S. workers in 1950 to an International Labour Organization estimate of 1.5% in 2025. The source describes work as shifting toward factories and services as farm productivity rose. That history provides context for changes in the kinds of jobs people do, but it does not show that AI will cause the same changes or at the same pace.
““Productivity isn’t everything, but in the long run it is almost everything.””
— Paul Krugman, quoted by Thorsten Meyer AI from a 1990 book
What AI Gains Show So Far
The scale of AI’s economy-wide effect is not settled in the source material. It reports an executive estimate of about 0.2 percentage points a year for AI’s contribution to productivity growth so far, but the excerpt does not specify the survey, period, sample or calculation behind that estimate. It also cites a 15% result for support agents without providing the study design or conditions needed to judge how broadly it applies.
The material does not establish how AI is affecting total employment, wages, working hours or job quality. Nor does it identify which industries have adopted the tools most widely or whether the gains will persist as organisations change their processes. The historical comparison with computers is suggestive context, not proof that AI will produce a similar productivity cycle.
Evidence to Watch in Coming Years
The clearest next indicator will be whether measured productivity growth rises as AI adoption spreads and firms reorganise work around the tools. Tracking that change requires specifying the measure and period: output per hour, for example, is different from the performance of one team or the time saved on one task.
Further evidence would also need to explain how the reported firm-level gains were measured, whether they hold across occupations and industries, and what happens to saved time and costs. Employment, wage and job-quality data will help show how any productivity gains are shared. Until those measures are available, the source supports a limited conclusion: some workplace gains have been reported, while AI’s broader effect remains uncertain.
Key Questions
What does productivity mean in this report?
Productivity means output per hour worked: the amount of value produced in an hour. It is an economy-wide measure when calculated across a whole sector or the economy, and can also be measured for a particular workplace or task.
What AI productivity gain does the article report?
The source cites a study reporting a 15% gain for support agents in one setting. It also reports an executive estimate that AI has contributed about 0.2 percentage points a year to productivity growth so far. The excerpt does not provide enough methodological detail to compare or generalise those figures.
Does the source show that AI is reducing jobs?
No. The provided material does not give data on AI’s net effect on employment, wages or job quality. It discusses productivity and historical shifts in the distribution of work, which do not by themselves establish what is happening to jobs now.
Why does the article compare AI with computers?
It uses the computer era to illustrate that widespread adoption and economy-wide productivity gains may not happen at the same time. The comparison is historical context; it does not establish that AI will follow the same timeline or produce the same results.
Source: Thorsten Meyer AI
