Within AI Bloom
What Would AI Abundance Actually Make Cheap?
Cheap, reproducible expertise could make many services abundant, but land, energy, materials and ownership would still limit who benefits.
On this page
- How intelligence becomes an economic input
- Which goods and services could approach abundance
- Why physical scarcity and ownership still matter
Page outline Jump by section
Introduction
Advanced AI could make skilled cognitive work radically cheaper without making the whole economy post-scarcity. Software-based expertise can be copied, run continuously and delivered across borders at very low marginal cost. That could expand access to tutoring, translation, design, diagnosis support, legal guidance, administration and technical problem-solving. Early workplace studies already show meaningful productivity gains in some tightly defined tasks.
But intelligence is only one input into human welfare. A perfect building plan does not supply bricks, land, electricity, machinery or planning permission. A medical recommendation does not create a hospital bed. Cheap educational content does not guarantee a quiet place to study. Even where AI produces valuable services almost freely, patents, platforms, professional rules and ownership structures can keep prices high.
The realistic prospect is therefore selective abundance: a world in which competent analysis and assistance become plentiful, while physical resources, trusted human responsibility and control over productive assets remain scarce. Whether that feels like abundance to ordinary people will depend as much on institutions and distribution as on AI capability.
How intelligence becomes an economic input
Most technologies make a particular physical process cheaper. Artificial intelligence is unusual because it can lower the cost of a general input: problem-solving.
Today, expertise is embodied mainly in people. Training a doctor, engineer, solicitor or teacher takes years. Each professional has limited working hours, cannot serve everyone at once and may be concentrated in wealthy cities or countries. Organisations therefore ration expertise through prices, queues, eligibility rules and simplified mass services.
A capable AI system changes this structure in three ways.
It is reproducible. Once developed, the same underlying model can assist many users. Additional use still requires computing power and electricity, but it does not require training another professional from childhood.
It is fast. AI can search, compare, calculate, translate, draft and explain in seconds. When reliable, this reduces the time needed for each case and allows human specialists to supervise more work.
It can absorb reusable knowledge. A useful procedure discovered in one organisation can potentially be incorporated into software and distributed rapidly, rather than spreading slowly through professional training and staff turnover.
Current systems offer an early, incomplete demonstration. In a study of 5,172 customer-support agents, access to an AI assistant increased the number of successfully resolved issues per hour by 15 per cent on average. Less experienced workers gained the most, suggesting that AI can partly reproduce practices previously held by the most capable staff.[OUP Academic]academic.oup.comOUP AcademicGenerative AI at Work* | The Quarterly Journal of Economics | Oxford AcademicFebruary 4, 2025 — Generative AI at Work* | The… Randomised field experiments involving 4,867 software developers at three companies found that access to a coding assistant increased completed tasks by about 26 per cent when the results were combined, although effects varied across teams and workers.[MIT Economics]economics.mit.eduMIT EconomicsThe Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software DevelopersThe Effects…
These results do not establish that expert labour is becoming free. They show a mechanism by which it could become less scarce: one experienced person’s methods can increasingly be embedded in a tool used by many less experienced people.
That matters because much of the modern economy is constrained not by a lack of raw information, but by the cost of applying it correctly. Regulations must be interpreted, symptoms assessed, lessons adapted, forms completed, designs checked and exceptions managed. If advanced AI can perform more of this work reliably, intelligence begins to resemble an inexpensive utility rather than a service rationed by professional time.
Which services could approach abundance?
The strongest candidates are services whose output can be delivered digitally and whose main cost comes from producing, interpreting or communicating information. Abundance here would not necessarily mean a price of zero. It would mean that high-quality assistance could be supplied to an additional person at a small fraction of today’s cost.
Advice, administration and routine analysis
Many organisations devote large amounts of labour to reading documents, preparing standard reports, reconciling records, answering recurring questions and guiding people through bureaucratic processes. Advanced AI could complete much of the first-pass work, leaving people to handle contested facts, unusual cases and final accountability.
This could reduce the cost of tax assistance, benefits applications, compliance checks, contract review, bookkeeping and basic business advice. Small firms and individuals who cannot presently afford specialists might gain access to capabilities once reserved for large organisations.
The economic value would come partly from lower fees and partly from avoided mistakes. A person who understands an entitlement, identifies a harmful contract clause or receives timely technical guidance may be better off even when no conventional product has been created.
Yet these services become abundant only where users are allowed to rely on them. Courts, regulators, insurers and professional bodies may still require a qualified person to sign, certify or accept liability. In high-stakes settings, trusted responsibility may remain scarcer than analysis itself.
Education and cognitive support
AI tutors can generate explanations, exercises, examples and feedback for each learner. Unlike a conventional textbook, a system can vary its vocabulary, pace and difficulty or translate material into a learner’s preferred language.
The potential source of abundance is not merely more content. The internet already contains more instructional material than anyone can consume. What remains scarce is responsive attention: noticing what a learner misunderstands and presenting the next explanation accordingly.
However, cheap tutoring software is not the same as universal education. Children still need motivation, safeguarding, social development and, in many cases, human teachers who can judge whether apparent progress is genuine. Evidence from programming education is already mixed: a 2026 meta-analysis found a moderate positive effect of generative AI tools on programming productivity, but no statistically significant improvement in learning outcomes across the included studies.[arXiv]arxiv.orgA meta-analysis of the effect of generative AI on productivity and learning in programmingMay 6, 2026…
AI may therefore make instruction abundant before it makes understanding abundant. The difference depends on assessment, pedagogy and whether learners remain mentally engaged rather than simply accepting generated answers.
Software, design and creative production
Software is especially susceptible because its inputs and outputs are largely digital. AI can help draft code, tests, documentation, interfaces and prototypes. Similar mechanisms apply to illustration, translation, video preparation, marketing materials and some forms of product design.
Falling production costs could allow small organisations to commission work that was previously uneconomic. A local charity might build a tailored database; a disabled user might generate an accessible interface; a scientist might create specialised analysis software without waiting for a dedicated development team.
But faster output can create new bottlenecks. Code must still be reviewed, integrated, secured and maintained. Creative material must be selected, legally cleared and matched to an audience. As generation becomes cheap, verification and judgement may become a larger share of total cost.
This is the “jagged frontier” observed in an experiment with 758 consultants. On tasks within the tested AI system’s capabilities, participants completed 12.2 per cent more tasks, worked 25.1 per cent faster and produced higher-quality answers. On a task outside that frontier, AI users were 19 per cent less likely to reach the correct result.[EconPapers]econpapers.repec.orgEconPapers: Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence… Abundant output is not the same as dependable expertise when users cannot tell which answers require correction.
Health and legal guidance
Medicine and law combine information-intensive analysis with physical, institutional and ethical constraints. AI could make symptom triage, literature review, record summarisation, standard legal explanations and second-opinion support much cheaper. That would be especially valuable where qualified professionals are scarce.
Even very capable systems, however, would not automatically make healthcare or justice abundant. Surgery requires facilities and staff. Medicines require manufacturing and supply chains. Legal outcomes depend on courts, enforcement and bargaining power. Many decisions also require an accountable person who can examine the patient, assess credibility or accept responsibility when something goes wrong.
In these fields, AI abundance is more likely to appear first as plentiful preliminary guidance and expanded professional capacity, not the disappearance of human institutions.
Why cheap intelligence does not remove physical scarcity
A useful way to test post-scarcity claims is to ask what happens after the AI has produced an answer.
Suppose an AI designs a low-cost home in seconds. Someone must still secure land, obtain permission, produce materials, transport them and build the structure. If land in a desirable city remains tightly limited, better architectural intelligence may improve the building without making housing affordable.
The same pattern applies elsewhere:
- An optimised electricity grid still needs cables, transformers and generating capacity.
- A newly discovered medicine still needs trials, factories, clinicians and distribution.
- An advanced agricultural plan still depends on soil, water, machinery and stable weather.
- A robotic production line still requires metals, chips, energy and maintenance.
- A personalised care plan still needs carers, equipment or accessible housing.
AI can help ease each bottleneck. It may design better batteries, reduce waste, improve logistics or accelerate construction methods. But solving a coordination or design problem does not abolish the underlying resource constraint.
AI itself is physical. Models run in data centres containing specialised chips, cooling systems and networking equipment. The International Energy Agency estimated that data centres used about 415 terawatt-hours of electricity in 2024, around 1.5 per cent of global electricity consumption, and projected demand of roughly 1,200 terawatt-hours by 2035 in its base case. It also identified grid connections and supplies of components such as transformers as constraints on expansion.[IEA]iea.orgExecutive summary – Energy and AI – Analysis - IEAwith its strong track record of identifying and exploring emerging issues in the ene… Its 2026 update reported that total data-centre electricity demand rose by 17 per cent during 2025 and warned that AI-focused facilities were encountering increasingly visible power and supply-chain bottlenecks.[IEA]iea.orgIf you prefer to log into your pData centre electricity use surged in 2025, even with tightening bottlenecks driving a scramble for solutions - News - IEAApril 16, 20…
Efficiency improvements may reduce the energy needed for each AI task, but cheaper use can increase total demand as more people use the technology for more ambitious purposes. Intelligence can therefore become more abundant while the infrastructure providing it remains capital-intensive and locally constrained.
This does not defeat the optimistic case. It changes its shape. AI abundance is most plausible when cheap intelligence is combined with abundant clean energy, automated production, improved recycling, faster construction and better infrastructure. Without those complements, civilisation may possess extraordinary designs and advice while struggling to implement them.
Ownership can preserve scarcity after costs fall
Prices do not simply reflect the amount of labour or material needed to produce something. They also reflect market power, legal rights and who controls access.
A digital service may cost very little to reproduce but remain expensive because one company controls the model, platform, data, distribution channel or required computing infrastructure. The history of software, pharmaceuticals and academic publishing shows that low copying costs do not automatically lead to low user prices.
AI infrastructure already has characteristics that could support concentration. An OECD review found high barriers to entry and relatively few suppliers across important parts of the compute chain, including advanced chips, cloud services and data-centre infrastructure. It also highlighted vertical integration, cross-investments and demand that frequently exceeds supply.[oecd.org]oecd.orgAt its core are the chip designers and manufacturers, who produce the essential hardware powering AI data centres. These firms rely on sp… A US Federal Trade Commission study of major cloud-provider partnerships with AI developers found equity, revenue-sharing, exclusivity and cloud-spending arrangements that may affect access to computing resources and increase switching costs.[Federal Trade Commission]ftc.govOpen source on ftc.gov.
This creates a distinction between technical abundance and economic abundance. A model may be capable of providing millions of inexpensive consultations, but the owner can still limit use, charge monopoly prices or reserve its strongest capabilities for favoured customers.
Other forms of ownership matter too. If AI makes companies vastly more productive while most households depend on wages, rising output need not translate into rising purchasing power. Economic models of transformative automation commonly find that automating a large share of production could accelerate growth while reducing labour’s share of income, especially if machines can substitute broadly for workers.[National Bureau of Economic Research]nber.orgOpen source on nber.org. The International Monetary Fund similarly warns that AI could raise capital returns and wealth inequality, even while sufficiently large productivity gains might raise incomes overall.[IMF eLibrary]elibrary.imf.orgOpen source on imf.org.
A society can therefore produce more and leave many people unable to buy it. This is not a technical contradiction. It is a distribution problem: productive capacity belongs to one group while purchasing power is concentrated elsewhere.
Work may become cheaper before life becomes cheaper
The transition to abundant intelligence is unlikely to occur all at once. Current evidence points towards task transformation more than immediate elimination of entire occupations.
The International Labour Organization’s 2025 assessment found that one in four workers worldwide were in occupations with some exposure to generative AI. Only 3.3 per cent of global employment fell into its highest-exposure category, and the organisation stressed that most affected jobs were more likely to be transformed than fully automated under current capabilities.[ilo.org]ilo.orggenerative ai and jobs 2025 updategenerative ai and jobs 2025 update
This matters for prices. When AI assists workers rather than replacing an entire production process, some savings may be absorbed by additional checking, software fees, organisational change or higher output quality. A solicitor may use AI to review more material but still charge for accountability. A teacher may prepare lessons faster but continue teaching the same number of pupils. A doctor may spend less time on documentation yet remain constrained by clinic space and appointment demand.
Higher productivity can also increase consumption instead of lowering expenditure. If design becomes cheaper, people may commission more designs. If diagnosis improves, health systems may identify more treatable conditions. If personalised education becomes widely available, learners may use more instructional support rather than spend less overall.
The post-scarcity question is therefore not simply whether the cost per unit falls. It is whether households gain reliable access to more of what they value: healthcare, housing, education, mobility, security and meaningful discretionary time.
What broad AI abundance would require
Technical progress is necessary, but several additional conditions determine whether cheap intelligence produces widely shared flourishing.
Reliable systems. AI must become dependable enough for users to know when it can act autonomously and when human review is essential. Otherwise, savings in generation may be offset by errors, verification and loss of trust.
Affordable compute and connectivity. Intelligence cannot be broadly available where devices, electricity, broadband or language support are missing. The IMF notes that lower-income economies may face less immediate disruption partly because they lack the infrastructure and skills needed to capture AI’s benefits, potentially widening differences between countries.[IMF eLibrary]elibrary.imf.orgOpen source on imf.org.
Competition and interoperability. Users need realistic alternatives between providers, the ability to move data and access to open standards. Public computing facilities, open models and shared research infrastructure may help prevent a handful of firms from becoming permanent toll collectors for machine intelligence. The OECD has identified public compute and support for open-source technology as possible ways to counter economies of scale and strengthen competition.[oecd.org]oecd.orgAt its core are the chip designers and manufacturers, who produce the essential hardware powering AI data centres. These firms rely on sp…
Access to physical complements. Housing, energy, transport, healthcare facilities and automated manufacturing must expand alongside cognitive capacity. Otherwise, AI may increase demand for scarce assets faster than supply, raising their prices.
A way to distribute purchasing power. If labour income falls relative to profits, abundance may require broader capital ownership, social dividends, public services, transfers, shorter working hours or other institutions that allow people to benefit from automated production. The correct mix is a political choice, not something an AI model determines.
Human freedom and recourse. Cheap advice is not flourishing when people cannot challenge decisions, choose providers or decline automated supervision. Systems that allocate benefits, medical treatment, credit or employment need transparent rules and routes to human appeal.
Post-scarcity is a distribution question
The phrase “post-scarcity” is most useful when treated as a direction rather than a literal destination. Advanced AI could make certain forms of expertise so inexpensive and reproducible that access to them ceases to be a major constraint. That would be historically important. Billions of people could gain practical capabilities that currently require money, geography, institutional status or years of specialist training.
But scarcity would migrate rather than disappear. As answers become plentiful, reliable verification may become scarce. As design becomes cheap, land and permission may become more valuable. As digital labour expands, energy and computing infrastructure may become bottlenecks. As production relies less on human work, ownership may matter more than wages.
The central question is therefore not whether AI can generate abundant intelligence. It increasingly appears capable of moving in that direction. The harder question is whether societies can connect that intelligence to energy, materials, institutions and rights in ways that reduce lived scarcity.
An AI-enabled bloom would not be measured by the number of generated reports, designs or recommendations. It would be visible when ordinary people can obtain good healthcare, effective education, secure housing, useful tools and greater control over their time without dependence on a narrow class of owners. Abundant intelligence is one possible foundation for that future. It is not, by itself, the finished structure.
Amazon book picks
Further Reading
Books and field guides related to What Would AI Abundance Actually Make Cheap?. Use these as the next step if you want deeper reading beyond the article.
AI Superpowers
THE NEW YORK TIMES, USA TODAY, AND WALL STREET JOURNAL BESTSELLER "Kai-Fu Lee believes China will be the next tech-innovation superpower...
The Zero Marginal Cost Society
First published 2014. Subjects: Cost, Cooperation, Capitalism, Coût, Costs.
The Age of Em
Robots may one day rule the world, but what is a robot-ruled Earth like? Many think the first truly smart robots will be brain emulations...
Abundance
Providing abundance is humanity's grandest challenge--this is a book about how we rise to meet it. Abundance for all is within mankind's...
eBay marketplace picks
Marketplace Samples
Live-tested eBay searches with available results related to this page.
Selected fromrobotics kit oneBay.co.uk.
Endnotes
1.
Source: academic.oup.com
Link:https://academic.oup.com/qje/article/140/2/889/7990658
Source snippet
OUP AcademicGenerative AI at Work* | The Quarterly Journal of Economics | Oxford AcademicFebruary 4, 2025 — Generative AI at Work* | The...
Published: February 4, 2025
2.
Source: economics.mit.edu
Link:https://economics.mit.edu/sites/default/files/inline-files/draft_copilot_experiments.pdf
Source snippet
MIT EconomicsThe Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software DevelopersThe Effects...
3.
Source: arxiv.org
Link:https://arxiv.org/abs/2605.04779
Source snippet
A meta-analysis of the effect of generative AI on productivity and learning in programmingMay 6, 2026...
Published: May 6, 2026
4.
Source: iea.org
Link:https://www.iea.org/reports/energy-and-ai/executive-summary?_hsenc=p2ANqtz-_XSDjYbWHnHsZzW0cp3y1v1eSDMqV8c73NX2yqdOFYyutAvsm0Ur5vBfMvwHwYZyOd5Iec
Source snippet
Executive summary – Energy and AI – Analysis - IEAwith its strong track record of identifying and exploring emerging issues in the ene...
5.
Source: iea.org
Title: If you prefer to log into your p
Link:https://www.iea.org/news/data-centre-electricity-use-surged-in-2025-even-with-tightening-bottlenecks-driving-a-scramble-for-solutions
Source snippet
Data centre electricity use surged in 2025, even with tightening bottlenecks driving a scramble for solutions - News - IEAApril 16, 20...
6.
Source: oecd.org
Link:https://www.oecd.org/en/publications/competition-in-artificial-intelligence-infrastructure_623d1874-en/full-report/component-3.html
Source snippet
At its core are the chip designers and manufacturers, who produce the essential hardware powering AI data centres. These firms rely on sp...
7.
Source: oecd.org
Link:https://www.oecd.org/en/publications/competition-in-artificial-intelligence-infrastructure_623d1874-en/full-report.html
8.
Source: elibrary.imf.org
Link:https://www.elibrary.imf.org/downloadpdf/view/journals/006/2024/001/006.2024.issue-001-en.pdf
9.
Source: imf.org
Link:https://www.imf.org/-/media/files/publications/sdn/2024/english/sdnea2024001.pdf
10.
Source: ilo.org
Title: generative ai and jobs 2025 update
Link:https://www.ilo.org/publications/generative-ai-and-jobs-2025-update
11.
Source: ilo.org
Title: Research brief FINAL 15May2025 21.05.25 1
Link:https://www.ilo.org/sites/default/files/2025-05/Research%20brief_FINAL_15May2025_21.05.25_1.pdf
12.
Source: policy.economy.ac
Title: ac The Energy Shock Behind Artificial Intelligence | The Economy
Link:https://policy.economy.ac/research/2026/05/202605289101
13.
Source: economy.ac
Title: The Energy Shock Behind Artificial Intelligence | The Economy
Link:https://economy.ac/research/2026/05/202605289101
14.
Source: elibrary.imf.org
Title: article A001 en.xml
Link:https://www.elibrary.imf.org/view/journals/068/2026/002/article-A001-en.xml
15.
Source: iea.org
Title: Key Questions on Energy and AI – Analysis
Link:https://www.iea.org/reports/key-questions-on-energy-and-ai
16.
Source: arxiv.org
Link:https://arxiv.org/pdf/2603.09209
17.
Source: arxiv.org
Link:https://arxiv.org/pdf/2603.09209v1
18.
Source: arxiv.org
Link:https://arxiv.org/abs/2603.09209v1
19.
Source: iea.org
Title: Energy and AI Observatory – Data Tools
Link:https://www.iea.org/data-and-statistics/data-tools/energy-and-ai-observatory?casestudy=AI+process+control+for+cement+production&tab=AI+for+energy
20.
Source: iea.org
Title: Energy and AI – Analysis
Link:https://www.iea.org/reports/energy-and-ai/
21.
Source: iea.org
Link:https://www.iea.org/news/ai-is-set-to-drive-surging-electricity-demand-from-data-centres-while-offering-the-potential-to-transform-how-the-energy-sector-works?_bhlid=0c3495641114dec7a09a90868d4994a9270380d9
22.
Source: iea.org
Title: global energy review 2025
Link:https://www.iea.org/reports/global-energy-review-2025?pubDate=20250329
23.
Source: elibrary.imf.org
Title: article A001 en.xml
Link:https://www.elibrary.imf.org/view/journals/001/2024/199/article-A001-en.xml
24.
Source: elibrary.imf.org
Title: article A001 en.xml
Link:https://www.elibrary.imf.org/view/journals/006/2024/002/article-A001-en.xml
25.
Source: elibrary.imf.org
Title: article A001 en.xml
Link:https://www.elibrary.imf.org/view/journals/006/2024/001/article-A001-en.xml
26.
Source: elibrary.imf.org
Title: article A000 en.xml
Link:https://www.elibrary.imf.org/view/journals/006/2024/001/article-A000-en.xml
27.
Source: arxiv.org
Link:https://arxiv.org/html/2603.09209
28.
Source: ar5iv.labs.arxiv.org
Link:https://ar5iv.labs.arxiv.org/html/2507.07019
29.
Source: arxiv.org
Link:https://arxiv.org/pdf/2304.11771v1
30.
Source: arxiv.org
Link:https://arxiv.org/html/2504.11436v3
31.
Source: elibrary.imf.org
Title: article A001 en
Link:https://www.elibrary.imf.org/downloadpdf/view/journals/006/2024/001/article-A001-en.pdf
32.
Source: imf.org
Title: wpiea2024116 print pdf
Link:https://www.imf.org/en/-/media/files/publications/wp/2024/english/wpiea2024116-print-pdf.pdf
33.
Source: imf.org
Link:https://www.imf.org/-/media/files/publications/sdn/2024/english/sdnea2024002.pdf
34.
Source: cef.imf.org
Title: 2024 05 07 Webinar on Gen AI Artificial Intelligence Slides
Link:https://www.cef.imf.org/content/dam/CEF-IMF/CEF-IMF-NewSite/3-News/1-CEFNewsandEvents/3-SeminarsWebinarsandHigh-levelEvents/2024-05-07%20-%20Webinar%20on%20Gen-AI%20Artificial%20Intelligence%20Slides.pdf
35.
Source: imf.org
Link:https://www.imf.org/-/media/Files/OAP/oap-home/2024/AISDNPPTMarch14.ashx?trk=public_post_comment-text
36.
Source: imf.org
Title: wpiea2024065 print pdf
Link:https://www.imf.org/-/media/files/publications/wp/2024/english/wpiea2024065-print-pdf.pdf
37.
Source: ilo.org
Title: Research brief GenAI 2025 Update
Link:https://www.ilo.org/sites/default/files/2025-05/Research%20brief_GenAI%202025%20Update.pdf
38.
Source: ilo.org
Title: WP140 web
Link:https://www.ilo.org/sites/default/files/2025-05/WP140_web.pdf
39.
Source: ilo.org
Title: intelligence artificielle generative et emploi revision 2025
Link:https://www.ilo.org/fr/publications/intelligence-artificielle-generative-et-emploi-revision-2025
40.
Source: ilo.org
Title: inteligencia artificial generativa y empleo edicion actualizada de 2025
Link:https://www.ilo.org/es/publications/inteligencia-artificial-generativa-y-empleo-edicion-actualizada-de-2025
41.
Source: ilo.org
Link:https://www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure
42.
Source: ilo.org
Link:https://www.ilo.org/publications/artificial-intelligence-adoption-and-its-impact-jobs
43.
Source: ilo.org
Link:https://www.ilo.org/sites/default/files/2025-07/ilo%20brief%20work%20transformed%20promise%20and%20peril%20of%20ai.pdf
44.
Source: labordoc.ilo.org
Link:https://labordoc.ilo.org/[discovery
45.
Source: ilo.org
Link:https://www.ilo.org/sites/default/files/2025-06/FWG_Note_Artificial%20intelligence%20adoption%20and%20its%20impact%20on%20jobs_ILO_2025_0.pdf
46.
Source: ilo.org
Title: G20 FWG AI adoption and its impact on jobs
Link:https://www.ilo.org/sites/default/files/2025-06/G20%20FWG_AI%20adoption%20and%20its%20impact%20on%20jobs.pdf
47.
Source: oecd.org
Title: component 4
Link:https://www.oecd.org/en/publications/competition-in-artificial-intelligence-infrastructure_623d1874-en/full-report/component-4.html
48.
Source: one.oecd.org
Link:https://one.oecd.org/document/DAF/COMP/M%282025%292/ANN4/FINAL/en/pdf
49.
Source: oecd.org
Link:https://www.oecd.org/en/publications/artificial-intelligence-markets_d531d73f-en/full-report.html
50.
Source: oecd.org
Link:https://www.oecd.org/en/publications/competition-in-the-age-of-ai_6f88a1ea-en.html
51.
Source: oecd.org
Link:https://www.oecd.org/en/publications/competition-in-artificial-intelligence-infrastructure_623d1874-en/full-report/component-5.html
52.
Source: oecd.org
Link:https://www.oecd.org/en/publications/competition-in-artificial-intelligence-infrastructure_623d1874-en.html
53.
Source: one.oecd.org
Link:https://one.oecd.org/document/DAF/COMP/M%282025%293/ANN2/FINAL/en/pdf
54.
Source: one.oecd.org
Link:https://one.oecd.org/document/DAF/COMP/M%282025%293/ANN2/FINAL/fr/pdf
55.
Source: one.oecd.org
Link:https://one.oecd.org/document/DAF/COMP%282025%2911/fr/pdf
56.
Source: oecd.org
Link:https://www.oecd.org/en/publications/competition-in-artificial-intelligence-infrastructure_623d1874-en/full-report/component-6.html
57.
Source: iea.org
Link:https://www.iea.org/reports/energy-and-ai/executive-summary?_hsenc=p2ANqtz–M5l5xvOfGlviNXajeGqNyI1_oeBo7FCL7ZTmZp0LPdOTcxGUmoUL6U5cJug2x-WpoyucPYpE_wJnxQ7xlKuXznUoDcQ&_hsmi=356121992
58.
Source: iea.org
Link:https://www.iea.org/reports/energy-and-ai/executive-summary%C2%A0
59.
Source: iea.org
Title: Energy demand from AI – Energy and AI – Analysis
Link:https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai?ref=explainx
60.
Source: iea.org
Title: electricity 2025
Link:https://www.iea.org/reports/electricity-2025
61.
Source: iea.org
Title: A I and climate change – Energy and AI – Analysis
Link:https://www.iea.org/reports/energy-and-ai/ai-and-climate-change?_hsenc=p2ANqtz-92nhAEJ6v6wAa9fOD6dJdaJY23XgWPjSVxKml4TRGyn6YRBD2FcrTUzrfTdT6GgUHEoMCIP-ufxbH35fBfYvu4qwZ1Fw
62.
Source: econpapers.repec.org
Link:https://econpapers.repec.org/RePEc%3Ainm%3Aororsc%3Av%3A37%3Ay%3A2026%3Ai%3A2%3Ap%3A403-423
Source snippet
EconPapers: Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of Artificial Intelligence...
63.
Source: ftc.gov
Link:https://www.ftc.gov/news-events/news/press-releases/2025/01/ftc-issues-staff-report-ai-partnerships-investments-study
64.
Source: nber.org
Link:https://www.nber.org/system/files/working_papers/w31815/w31815.pdf
65.
Source: ftc.gov
Link:https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/2025/01/behind-ftcs-6b-report-large-ai-partnerships-investments
66.
Source: ftc.gov
Link:https://www.ftc.gov/legal-library/browse/joint-statement-competition-generative-ai-foundation-models-ai-products
67.
Source: ftc.gov
Title: Generative AI Raises Competition Concerns | Federal Trade Commission
Link:https://www.ftc.gov/policy/advocacy-research/tech-at-ftc/2023/06/generative-ai-raises-competition-concerns
68.
Source: search.ftc.gov
Title: ai joint statement
Link:https://search.ftc.gov/system/files/ftc_gov/pdf/ai-joint-statement.pdf
69.
Source: ftc.gov
Link:https://www.ftc.gov/system/files/ftc_gov/pdf/ai-joint-statement.pdf
70.
Source: search.ftc.gov
Title: otstaffreport techaisummitquotebook
Link:https://search.ftc.gov/system/files/ftc_gov/pdf/ftcotstaffreport-techaisummitquotebook.pdf
71.
Source: ftc.gov
Link:https://www.ftc.gov/system/files/ftc_gov/pdf/ftcotstaffreport-techaisummitquotebook.pdf
72.
Source: ftc.gov
Link:https://www.ftc.gov/reports/ftc-staff-report-ai-partnerships-investments-6b-study
73.
Source: ftc.gov
Link:https://www.ftc.gov/system/files/ftc_gov/pdf/p246201_aipartnerships6breport_redacted_0.pdf
74.
Source: ftc.gov
Link:https://www.ftc.gov/system/files/ftc_gov/pdf/12-15-2023AICEStaffReport.pdf
75.
Source: search.ftc.gov
Title: cloud computing taking stock and looking ahead may 11 2023
Link:https://search.ftc.gov/system/files/ftc_gov/pdf/cloud-computing-taking-stock-and-looking-ahead-may-11-2023.pdf
76.
Source: nber.org
Link:https://www.nber.org/system/files/working_papers/w31815/revisions/w31815.rev1.pdf
77.
Source: nber.org
Link:https://www.nber.org/system/files/working_papers/w31815/revisions/w31815.rev2.pdf
78.
Source: nber.org
Link:https://www.nber.org/papers/w31815
79.
Source: nber.org
Link:https://www.nber.org/system/files/working_papers/w31815/revisions/w31815.rev0.pdf
80.
Source: nber.org
Link:https://www.nber.org/papers/w31161
81.
Source: nber.org
Link:https://www.nber.org/system/files/chapters/c15298/c15298.pdf
82.
Source: nber.org
Link:https://www.nber.org/papers/w34423
83.
Source: ideas.repec.org
Title: v37y2026i2p403 423
Link:https://ideas.repec.org/a/inm/ororsc/v37y2026i2p403-423.html
Additional References
84.
Source: nature.com
Link:https://www.nature.com/articles/s41746-025-01722-y
85.
Source: youtube.com
Title: Making Sense of Post Labor and Post Scarcity Economics
Link:https://www.youtube.com/watch?v=89VObIY0A2s
Source snippet
Is AI Productivity Worth Our Humanity? with Prof. Michael Sandel...
86.
Source: youtube.com
Title: Is AI Productivity Worth Our Humanity? with Prof. Michael Sandel
Link:https://www.youtube.com/watch?v=L1U2haTj-Ws
Source snippet
Can AI supercharge global economic growth?...
87.
Source: youtube.com
Title: Raj Reddy: The Future of AI: Doomers vs. Abundance
Link:https://www.youtube.com/watch?v=ydnOSMbyyQo
Source snippet
Making Sense of Post Labor and Post Scarcity Economics...
88.
Source: academia.edu
Link:https://www.academia.edu/165349603/Constraints_and_Systemic_Risks_in_a_Robot_Saturated_Post_Scarcity_Economy_An_Institutional_and_Political_Economy_Analysis_of_Technological_Transition_Dynamics
89.
Source: artificialgeniusintelligence.com
Link:https://artificialgeniusintelligence.com/assets/papers/Age_of_Abundance_Post-Labor_World.pdf
90.
Source: brookings.edu
Link:https://www.brookings.edu/projects/artificial-intelligence-and-emerging-technology-initiative/
91.
Source: dunham.com
Link:https://dunham.com/FA/Blog/Posts/why-healthcare-and-education-costs-keep-rising
92.
Source: globalprioritiesinstitute.org
Link:https://www.globalprioritiesinstitute.org/wp-content/uploads/Economic-growth-under-transformative-ai-Philip-Trammell-and-Anton-Korinek.pdf
93.
Source: brookings.edu
Link:https://www.brookings.edu/articles/from-open-internet-to-open-intelligence-why-ais-market-structure-matters-more-than-ever/