TL;DR
An opinion piece argues that broad replacement of human workers with AI risks hollowing out the payroll tax base that funds public services. The author urges keeping humans in supervisory roles and suggests taxing value flows or data-center energy rather than treating AI as a taxable person.
What happened
The author recounts overhearing a worker in River Falls, Wisconsin, struggle with an AI feature (likely Microsoft Copilot) in a spreadsheet, ultimately abandoning the tool and doing the job manually. That episode prompts a broader claim: when AI replaces paid labor, the government loses payroll tax revenue that funds schools, roads and healthcare. The piece challenges common counterarguments — that corporations can cover the gap with profit taxes, that AI is just another tool like tractors, or that software cannot be taxed because it lacks a physical location. The author contends those responses are inadequate, citing corporate tax avoidance, the rapid tempo of AI-driven displacement, and the difference between mechanizing muscle and automating cognitive tasks. Proposed alternatives include taxing the value generated domestically or the energy footprint of data centers, and preserving human-in-the-loop arrangements to maintain both quality control and the tax base.
Why it matters
- Loss of payroll tax revenue could reduce funding for public infrastructure and services.
- Relying on corporate profit taxes may be insufficient because profits can be shifted or masked.
- Rapid AI displacement differs from past technological shifts and may outpace retraining efforts.
- Policy choices on how to tax AI-driven value will shape who bears the costs of automation.
Key facts
- Anecdote: a worker in River Falls, Wisconsin, struggled with an AI spreadsheet feature (likely Microsoft Copilot) and completed the task manually.
- Central claim: replacing paid human labor with AI removes a source of payroll tax revenue that funds public services.
- Argument against corporate-tax-only fixes: corporations can minimize reported profits through IP shifting, buybacks, or reinvestment strategies.
- Comparison: tractors amplified physical labor over decades; AI replaces cognitive work much faster, in quarters rather than decades.
- The author recommends maintaining humans as quality controllers (human-in-the-loop) to sustain tax contributions and output quality.
- Suggested tax targets include value flows (revenue generated in-country) or the energy consumption of data centers, rather than granting AI personhood.
- The piece asserts that if roughly 30% of the workforce were displaced, the remaining workers could face an unsustainable burden to fund social support — presented as a mathematical concern.
- The author expresses optimism about large language models for certain work (notably software engineering) but warns enthusiasm should account for economic consequences.
What to watch next
- Policy proposals that target taxation of revenue generated within a country or of data-center energy use rather than taxing software directly.
- Rates and pace of workforce displacement by AI tools across industries, since rapid change may outstrip retraining efforts.
- not confirmed in the source
- not confirmed in the source
Quick glossary
- Human-in-the-loop: A workflow design where human judgment or oversight remains part of a process that also uses automated systems.
- Payroll tax: Taxes collected from wages, often funding social insurance programs and other public services.
- Large language model (LLM): A type of AI trained on large text datasets to generate or analyze human-like language; used in assistants and code tools.
- Corporate profit tax: A tax on a company's reported profits; its effectiveness can be affected by accounting choices and international tax strategies.
- Value flow taxation: A tax approach aimed at capturing the economic value generated in a jurisdiction, regardless of where software or IP is hosted.
Reader FAQ
Can we simply tax companies' higher profits to replace lost payroll taxes?
The source argues no — it says companies can minimize reported profits through tactics like IP shifting, buybacks, or showing reinvestment, making profit taxes an unreliable replacement.
Is it feasible to tax AI itself as a legal entity?
The source states you can't tax the code or give software personhood to solve the problem; instead it suggests taxing value flows or data-center energy usage.
Will retraining displaced workers solve the tax-base problem?
The author contends retraining is unlikely to keep pace with rapid AI-driven displacement and calls the idea a myth in that context.
Are investors looking to avoid societal costs from automation?
The piece asserts some investors prioritize extracting returns and exiting, but this is an opinion presented by the author rather than independently verified in the source.
AI Employees Don't Pay Taxes DECEMBER 29, 2025 I recently overheard a conversation in a River Falls, Wisconsin, coffee shop that has stayed with me. A person was venting to…
Sources
- AI Employees Don't Pay Taxes
- How AI Will Reshape Employment Taxes And Public …
- Inclusive Growth in the Era of Automation and AI: How Can …
- Should Robots Pay Taxes? Tax Policy in the Age of …
Related posts
- USPS finalizes DMM rule redefining postmark dates and possession
- Why the Internet Is Bad for Democracy – A 2005 ACM Paper Reviewed
- USPS Finalizes New Rule Defining Postmarks and Postal Possession Dates