🔍 Read the full analysis: What AI Metrics Say About Capability And Performance on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
A November 11, 2026, ThorstenMeyerAI.com report argues that metrics for economic output and AI capability can show real changes while missing important parts of their value. It cites estimates for unpaid work and consumer benefits from AI chatbots, while warning that the reported figures do not by themselves establish broad welfare gains or AI performance across tasks.
ThorstenMeyerAI.com published a report on November 11, 2026, arguing that familiar economic measures and newer AI estimates capture only selected parts of what they seek to describe. The report points to a Stanford working paper estimating $172 billion a year in value to U.S. consumers from AI chatbots, while cautioning through the history of national accounting that measured output is not the same as overall welfare or a complete account of capability.
The report begins with the creation of national income accounting during the Great Depression. On January 4, 1934, economist Simon Kuznets’ report to the U.S. Senate estimated that income produced in the United States fell from $83.0 billion in 1929 to $39.4 billion in 1932, a decline of 53%. The figures gave policymakers a way to quantify the downturn, but Kuznets warned that national income could not, on its own, establish how well a nation was doing.
The article uses that distinction to frame current claims about AI. It cites a Stanford working paper that estimates U.S. consumers receive $172 billion in annual value from AI chatbots, an amount the report says is far above the revenue those tools earn. It also describes a 2017 experiment in which the median participant would have required $17,530 to give up search engines for one year, despite paying no direct price to use them. These figures illustrate a gap between consumer value and market revenue; they are not direct measures of a model’s accuracy, reasoning ability or usefulness in every setting.
The report also gives a measure of work that GDP commonly excludes. In Germany, unpaid household work was valued at €987 billion in 2013 using a wage-based approach, equivalent to 39% of the value added counted in GDP, according to the source material. The example reinforces the report’s central point: a metric can be useful for a defined purpose while leaving substantial activity outside its frame.
Infographic 01 · The sector clock, by output
China: how GDP changed shape
Share of GDP (value added at current prices) by sector, 1978 to 2025. In 2023 farming still held 22.8% of all jobs but produced 7.1% of GDP.
- Farming27.7%
- Industry47.7%
- Services24.6%
Reform starts
- Farming14.7%
- Industry45.5%
- Services39.8%
- Farming9.1%
- Industry45.4%
- Services45.5%
Services overtake industry
- Farming6.7%
- Industry35.6%
- Services57.7%
Communiqué 2025
Measured by output, not by jobs: the sector clocks of earlier episodes count workers. Jobs vs. output for farming, 2023: 22.8% of employment (year-end) vs. 7.1% of GDP.
Sources: NBS China, China Statistical Yearbook 2024, tables 3-2 and 4-2; NBS, Statistical Communiqué 2025 (28 Feb 2026).
Powered by Thorsten Meyer AI
Why AI Value Outruns Revenue
The distinction matters as governments, companies and consumers try to judge AI’s effects. Revenue, usage, benchmark scores and consumer surplus measure different things. A chatbot may deliver value to users without capturing that full value as sales, just as an economic account can omit unpaid services. Conversely, a large estimate of consumer benefit does not prove that every user benefits equally or that a system performs reliably across tasks.
For readers evaluating AI claims, the report’s examples point to a practical need to ask what a metric counts, who produced it and what comparison it uses. A capability score is not automatically a measure of real-world performance; results depend on the tasks, test conditions and population being measured. The material supplied does not provide benchmark results or a detailed methodology for the chatbot estimate, so the $172 billion figure should be read as a working-paper estimate rather than an established economy-wide outcome.
Products Worth Considering
Samsung Odyssey G9 QD-OLED 49 Dual QHD Curved Gaming Monitor
Apple Watch Series 8 GPS 41mm Midnight Aluminum Case with Midnight Sport Band
Bose TV Speaker Compact Soundbar with Bluetooth and HDMI-ARC, Black
As an affiliate, we earn on qualifying purchases.
From Kuznets to AI Measures
The U.S. government’s early national income estimates were produced in response to a specific policy need: understanding the scale of the Depression. In the 1934 report, Kuznets said the estimates primarily covered activities whose results appeared in markets. He warned that they left out much household work and did not show how income was distributed or how difficult the work was. His caution, quoted in the source, was that “the welfare of a nation can, therefore, scarcely be inferred” from national income alone.
National accounting later developed into measures of total output. The source traces gross national product estimates in the 1940s, the first international System of National Accounts in 1953, and the United States’ adoption of GDP as its headline measure in 1991. GDP counts the market value of final goods and services produced within a country. That definition makes it useful for tracking economic activity, but it does not make GDP a full measure of welfare. The report applies the same caution to AI: what can be counted depends on the metric’s definition.
The source also notes that AI tools may create value for users that is not reflected in the revenue they generate. This is a familiar measurement problem, not proof that any particular AI system is broadly capable. The supplied material contains no independent evaluation of chatbot accuracy, safety or performance across specific benchmarks.
Limits of the AI Estimate
The source material does not identify the Stanford working paper by title, authors, publication date or methods. It also does not explain how the $172 billion annual estimate was calculated, what counts as an AI chatbot, or how uncertainty and differences among users were handled. The figure should not be treated as a settled measurement until those details are available.
No specific AI benchmark scores, model comparisons or test results are supplied. It is therefore unclear from this material how the article’s broader discussion of metrics relates to the performance of any particular system. The reported consumer-value estimate also does not establish how benefits compare with costs, risks or outcomes across different groups.
Method Details Will Matter
The next step for readers is to examine the underlying Stanford working paper and its methodology, including the population studied, the time period, the definition of consumer value and the uncertainty range. Those details would show how far the estimate can be generalized beyond the study.
More broadly, claims about AI capability will be easier to assess when reports identify the tasks tested, the comparison baseline, the evaluation conditions and the limits of the results. Revenue, consumer benefit and benchmark performance should remain distinct measures unless evidence supports combining them. The source material does not specify a forthcoming release or milestone for the working paper, so no publication schedule is confirmed.
Key Questions
What is the main news in the report?
ThorstenMeyerAI.com published a report on November 11, 2026, linking the history of GDP’s limitations to modern efforts to measure AI capability and value. It cites a working-paper estimate of consumer value from AI chatbots while emphasizing that such estimates measure only a defined part of AI’s effects.
Does the $172 billion figure measure AI performance?
No. The source describes it as an estimate of annual value to U.S. consumers from AI chatbots. It does not provide benchmark scores or establish accuracy, reliability or performance across tasks.
Why does the report discuss GDP and unpaid work?
The comparison shows how a widely used metric can be useful while omitting activity outside its definition. GDP primarily counts market production, while unpaid household work is generally not included in the same way.
What is still unknown about the chatbot estimate?
The supplied material does not name the working paper or describe its methods, sample, uncertainty range or definition of consumer value. Those details are needed to judge how broadly the $172 billion estimate applies.
Source: ThorstenMeyerAI.com
