AI and Jobs: What a New Survey Actually Shows

The Question

Nearly four years after ChatGPT's launch, is AI actually taking jobs, creating them, or both? A sociologist researching AI decided to find out by going straight to workers instead of relying on speculation or employer-side data.

The Study

Who: YouGov surveyed 1,250 employed U.S. adults, weighted to reflect the national workforce (age, gender, race, education).
When: July 30 – August 4, 2026.
What: A 25-question survey asking workers directly whether AI had changed their job status since 2023 — through job loss, new job creation, or promotion.
Status: Part of an ongoing study; results are not yet peer-reviewed.

The Three Core Questions

Have you lost a job to AI automation since 2023?
Have you landed a brand-new job that didn't exist before AI (e.g., AI data labeler, AI compliance specialist)?
Have you gotten a promotion you credit to AI, such as AI-related skills?

The Results

Outcome	Share of Workers
Lost a job due to AI	~3%
Landed a new, AI-created job	~6%
Got a promotion tied to AI	~9%
"Not sure" (per question)	3–4%

Roughly 88–95% of respondents said no to each individual question — meaning AI-driven job change of any kind remains rare so far.

The Takeaway

No mass disruption yet. The data doesn't support either the doom scenario (AI wiping out jobs) or the boom scenario (AI creating a wave of new roles) — at least not yet, and not at scale.
Promotions are the biggest signal. Of the three effects measured, career advancement tied to AI skills was the most common — suggesting AI may currently reward those who adopt it rather than replace those who don't.
Context matters. Job turnover in the U.S. is generally low, so even these small percentages should be read against a backdrop of general job-market stability, not as a null result.
Agentic AI is the wildcard. More autonomous AI systems could make today's "AI skills" less valuable and accelerate job displacement in ways this snapshot can't capture.

Caveats Worth Noting

The survey captures a single point in time (July 2026), when unemployment was 4.1%, but the workforce was already shrinking, with AI cited as a factor in layoffs elsewhere.
Respondents were opt-in panel members, not randomly selected — and all were currently employed, so anyone who lost a job to AI and remains jobless isn't represented.
Workers may not always know for certain whether AI specifically caused their job change.

Why It's Different

Unlike trackers that focus only on AI-related layoffs, this survey also measures job creation and promotion — giving a fuller, if still early, picture of how AI is showing up in workers' actual careers.