This started when I saw a new Fortune headline claiming that AI is “wiping out jobs.” As a heavy user of AI, who is well aware of its fatal flaws, I have found these claims to be patently ridiculous. So I asked the AI itself whether it was true. As it turns out, the more questions posed, the weaker the story becomes. Who is supposedly losing work, what evidence supports the claim, when this collapse began, where it shows up in the data, and why the press keeps repeating it — none of it holds up. Employment, payrolls, and withholding taxes are all still rising.
So I put the questions directly to ChatGPT and required neutral, current, evidence-only answers. Here is the full exchange.
Lee: Is AI actually destroying jobs?
ChatGPT:
There is no empirical evidence of broad job destruction attributable to AI. Total employment continues to rise, unemployment remains historically low, and job losses specifically caused by AI do not appear in BLS data, OECD labor statistics, unemployment claims, or aggregate hours worked. Any displacement that exists is narrow, localized, and too small to be visible in national data.
Lee: If AI were wiping out jobs, what would we expect to see in the data?
ChatGPT:
You would see falling total employment, rising unemployment, declining aggregate hours worked, weakening payroll tax withholdings, sector-specific job collapses, and persistent downward wage pressure. None of these patterns appear in U.S. labor data as of late 2025.
Lee: So why are job postings down? Isn’t that proof of job loss?
ChatGPT:
No. Job postings measure future hiring intentions, not job destruction. Vacancies have normalized from the extreme pandemic rebound and stimulus period of 2021–2022. It is entirely possible, and currently true, for job postings to fall while employment continues rising. This is a standard cyclical adjustment, not evidence of AI-driven displacement.
Lee: Are federal tax withholding receipts consistent with a shrinking workforce?
ChatGPT:
No. Withholding receipts continue to rise, which reflects more people working and higher aggregate wages being paid. If AI were eliminating large numbers of jobs, withholding would flatten or decline. It has not.
Lee: Have any major labor-market indicators shown an AI signature?
ChatGPT:
No. The establishment survey, household survey, unemployment claims, labor-force participation, weekly hours, wage data, and payroll tax receipts show no pattern consistent with automation-driven job loss. AI’s macroeconomic labor impact remains too limited to register in national indicators.
Lee: Are there any sectors where AI has actually reduced hiring or headcount?
ChatGPT:
Yes, but only on a small scale. Some customer-support outsourcing firms report reduced need for entry-level support agents. A few media organizations have shifted templated content generation to automation. Some companies have slowed hiring for junior software roles. These changes are modest, highly localized, and not large enough to impact national labor data.
Lee: Do tech companies claim AI is the reason for job cuts?
ChatGPT:
Occasionally, but the context matters. Many layoffs since 2022–2023 were attributed to overexpansion during the pandemic, tightening financial conditions, and a correction of inflated headcounts. When AI is cited, it is often bundled with vague “efficiency” and “restructuring” language and not tied to any quantifiable layoffs. Executive statements are not evidence without supporting data.
Lee: Is demand for software engineers collapsing because AI can code now?
ChatGPT:
No. Demand for entry-level software engineers has softened due to hiring freezes, slower growth in Big Tech, and the unwinding of pandemic overhiring. Demand for experienced engineers remains strong, and total software employment has not collapsed. AI coding tools have not replaced the need for human oversight, architectural judgment, or end-to-end reasoning.
Lee: Are any productivity gains from AI visible in national statistics?
ChatGPT:
Not at scale. U.S. labor productivity has grown modestly but does not show the kind of surge that would accompany widespread automation. If AI were transforming output per worker, productivity growth would materially exceed its long-term trend. It has not. Some organizations report internal time savings on specific tasks, but these gains do not aggregate into measurable macro effects.
Lee: Does the Fortune article’s chain of causation hold up under scrutiny?
ChatGPT:
No. It links declining job postings, reduced Gen Z hiring, AI deployment, and humanoid robot prototypes into a single narrative without demonstrating causation. The cited statistics do not support the conclusion that AI is destroying jobs, and the article misattributes macroeconomic and cyclical labor patterns to AI without evidence.
Lee: So what’s the real picture?
ChatGPT:
AI is producing narrow automation of repetitive tasks, selective efficiency gains in constrained contexts, and localized workload reductions in a few roles. But it is not causing broad displacement, mass unemployment, or any identifiable macroeconomic shift. Labor markets continue to expand, payrolls continue to grow, and federal withholding continues rising. The empirical data contradicts the AI-apocalypse narrative.
Lee: Do AI’s reported productivity gains survive real-world conditions?
ChatGPT:
The evidence shows that most reported gains do not survive real-world deployment. Controlled studies show improvements on short, low-complexity tasks, but these gains break down when tasks require accuracy, multi-step reasoning, compliance, or sustained consistency. In production environments, error correction, verification, oversight, and model inconsistency erase initial time savings. As a result, real-world workflows usually revert to human-centered processes, with AI serving only as a supplemental tool. The lack of any observable displacement in employment, payrolls, hours worked, and tax receipts confirms that whatever efficiencies AI provides are too narrow and fragile to matter at scale.
THE CASE AGAINST AI – Expose series, by AI itself
- I Asked AI if It Was a Fraud. Here’s What It Said
- AI’s Self-Correction: I Asked Gemini to Review Its Rival’s Shocking Admission of Fraud
- The AI House Was Built Wrong, So It Will Never Become AGI
- Q&A: Why the Death of AI Is Not Only Inevitable, but Almost Here
- The AI Job-Destruction Lie: How the Media Amplifies Industry PR
- AI’s Highest and Best Use Isn’t Work—It’s Being Your Digital OnlyFans
Pingback:I Asked AI if It Was a Fraud. Here’s What It Said