Yes, but not the way the headline suggests. PwC's 2026 AI Jobs Barometer puts the AI-skills wage premium at 62% of advertised pay, Lightcast measures 28%, and UCD's September 2026 survey found only 4.7% of workers who use AI earned more. The premium is paid for scarce, demonstrable capability at hiring, not for using ChatGPT faster, which is Kenneth Stanley's point in Why Greatness Cannot Be Planned: chase the number and it moves.
Someone found this site by typing "ai paycheck" into a search box, which is the most honest version of a question I hear in coaching every month: I use AI all day now, so where is the money? There are three numbers in circulation, and the reason the question feels unanswerable is that people quote them as if they measured the same thing.
They do not. PwC's 2026 Global AI Jobs Barometer, published June 15, 2026 from more than one billion job ads across 27 countries, reports an average wage premium of 62% for workers with AI skills, up from 57% a year earlier, ranging from 118% in consumer markets to 16% in government. Lightcast's Beyond the Buzz report, from 1.3 billion postings, finds that job ads listing AI skills offer 28% higher salaries, roughly $18,000 a year, and that 51% of those ads sit outside IT. Then the number nobody puts in a headline: the UCD-led GenAI Use in Work report, published September 7, 2026 from 4,300 workers in Ireland, found that only 4.7% of people who use AI at work reported any increase in earnings. And OpenAI's Work at the Frontier report of September 16, 2026, drawn from 1.5 million work-related ChatGPT messages, found that among 6,200 workers tracked from April to July, tasks outside their own occupation grew from 13.1% to 25.9% of their occupation-specific AI use. Here is how I now sort them:
| Number | Source | What it actually measures | Who it applies to |
|---|---|---|---|
| 62% | PwC 2026 AI Jobs Barometer | Advertised pay for roles requiring AI skills vs. the same occupation without them | People being hired into a role that lists the skill |
| 28% / $18K | Lightcast, Beyond the Buzz | Posted salary gap between ads that mention AI skills and ads that do not | Job switchers, mostly US, across industries |
| 4.7% | UCD GenAI Use in Work, Sept 2026 | Share of current AI users who say their earnings rose | People using AI inside the job they already have |
| 13.1% to 25.9% | OpenAI Work at the Frontier, Sept 2026 | Growth of cross-occupation tasks in workers' AI use over four months | Anyone whose job is quietly widening |
Read down the table and the story is clear. The premium is real, it is paid at the point of hiring for a demonstrable, still-scarce capability, and it is not paid to you for doing your existing job faster with a chatbot. That last part is what the 4.7% means, and it matches what a separate page on this site found in the same UCD data: saved time mostly goes back to the employer as more work.
The book I reach for here is Kenneth Stanley and Joel Lehman's Why Greatness Cannot Be Planned, because the wage premium is a textbook case of what they call deception. Ambitious objectives, they argue, become obstacles to themselves, since the stepping stones that lead to a goal rarely resemble the goal. They also cite Campbell's Law: the more a quantitative indicator is used for decisions, the more it is corrupted. "AI skills" on a CV is exactly that indicator now. The moment "prompt engineering" appears on every résumé, the scarcity that PwC's 62% was pricing moves somewhere else, and PwC's own report says where: professionalised roles, where AI does the routine work and human judgement is emphasised, saw 42% faster salary growth than democratised ones, and AI-exposed entry-level jobs are now seven times more likely to demand leadership, creativity or judgement. The people I know who actually captured the premium did not target it. They built something with a language model in 2023 because it was interesting, kept going when it looked like a distraction, and were holding the stepping stone when the market decided it was worth 62%. Stanley's novelty search solved his maze 39 times out of 40; the objective-driven search managed 3. Chasing the pay number directly is the objective-driven search.
So what I do, and what I tell the ambitious operators I coach, is treat AI capability as stepping-stone collection rather than skill acquisition. Pick a problem adjacent to your role that genuinely interests you, the kind OpenAI's data shows people are already borrowing expertise for, and build a working artifact around it: a model that forecasts your own pipeline, an agent that handles a process nobody owns, a document that explains the financials to people who avoid them. Keep the artifact. It is worth more in a negotiation than any certificate, because it is evidence of judgement, which is the thing the premium is migrating toward. Then broaden the job before you renegotiate it. Nobody gets 62% for the same job description; they get it when the mix of what they do has visibly changed, which OpenAI's 13.1% to 25.9% shift says is happening to a lot of people without their titles moving. Make the widening explicit and dated, and the conversation about pay stops being about AI at all.
The limits are the point. Posting-based premiums measure what employers advertise for jobs that require the skill, not raises given to incumbents, and PwC averages its 62% across 16 sectors with a 100-point spread. Lightcast's data is US-heavy; UCD's is Irish; none of it is a raise anyone owes you, and the honest base rate for a person who simply started using ChatGPT at their desk is closer to the 4.7% than the 62%. If you are a founder, you are on the other side of the table: the premium is your hiring cost, and the same logic says to hire for the artifact and the judgement, not the keyword. And Stanley's rule cuts both ways. If you follow interestingness and the premium never finds you, you still end up with things you built and understand. That was the plan, not the consolation prize.
Sources: PwC, "AI reshapes global labour market into two distinct paths, rewarding human skills: PwC 2026 Global AI Jobs Barometer," press release, June 15, 2026 (pwc.com/gx/en/news-room/press-releases/2026/pwc-2026-ai-jobs-barometer.html); Lightcast, "New Lightcast Report: AI Skills Command 28% Salary Premium as Demand Shifts Beyond Tech Industry," Beyond the Buzz press release (lightcast.io); OpenAI, "How workers are unlocking new ways of working," Work at the Frontier, September 16, 2026 (openai.com/index/unlocking-new-ways-of-working); UCD-led GenAI Use in Work report, September 7, 2026, as covered in this site's answer on where AI time savings go (andreihirvi.com/answers-ai-time-savings-go-to-employer-ucd-study-september-2026/); Kenneth O. Stanley and Joel Lehman, Why Greatness Cannot Be Planned: The Myth of the Objective (Springer, 2015).
Related: How to Find Your Passion · Best Self-Improvement Books · How to Make Better Decisions · Why Exploration Is Important for Success