The judgement premium is the extra value that attaches to work requiring a person to decide what a situation is, once the parts of the job that could be specified in advance are cheap. It is visible in what employers now ask for. It has not been measured in what they actually pay for judgement itself, and this page keeps those two things apart.
The answer, in one line
Definition#
The judgement premium: the additional value that accrues to roles where a human still decides what matters, as automation drives down the price of the specifiable work around them.
Two tracks, and only one of them pays more#
PwC's 2026 Global AI Jobs Barometer, published 15 June 2026 on more than a billion job advertisements across six continents, describes a labour market splitting in two. In roles PwC call professionalised, AI takes the routine tasks and the advertisement leans harder on human judgement and expertise. In roles they call democratised, AI makes the job itself easier for a non-expert to do. Radiologists and recruiters sit on the first track, IT service managers and medical secretaries on the second.
Professionalised roles show twice the growth in available jobs and 42 per cent faster salary growth. The entry-level finding is sharper still. Across 2.4 million US entry-level advertisements, the roles most exposed to AI are seven times more likely to require traditionally senior human-intensive skills such as leadership, creativity or face-to-face interaction, and those roles grew 35 per cent since 2019 while other entry-level roles fell 10 per cent. Graded entry.
Two cautions belong with the figure. This is stated employer demand rather than realised pay: an advertisement is a wish, and the salary growth PwC report is advertised salary. And PwC sells services into the market it is measuring. The separate 62 per cent wage premium in the same report attaches to AI skills, which is a different quantity and often quoted as though it were this one.
The direction test underneath it#
The economic reason to expect a premium at all comes from Autor and Thompson, published in the Journal of the European Economic Association after circulating as NBER Working Paper 33941. Across four decades of task data covering 303 US occupations, they find that what matters is which tasks a technology removes. Automation that stripped out the less expert tasks raised wages and reduced employment in the occupation. Automation that stripped out the expert tasks lowered wages and raised employment. Graded entry.
That gives the premium a testable form. Ask of any role whether the system is taking the preparation or taking the decision. Where it takes the preparation, the remaining work concentrates into judgement and the role appreciates. Where it takes the decision, the person is left assembling inputs for a machine and the role commoditises. Their data ends in 2018, so this is a lens for reading the current evidence and not a measurement of generative AI.
Nobody has priced judgement itself#
No study on this estate measures a wage premium for judgement as a capability. What exists is a premium for AI skills, a growth and salary gap between two categories of role, and a historical result about the direction of automation. Assembling those into a number would be inventing one.
There is also a live counter-current. The compression results collected on what becomes more valuable as AI gets cheaper show AI narrowing the gap between the strongest and weakest performers on the tasks it does well, which is a mechanism that would erode a premium rather than build one. Both can be true at once if they apply to different tasks. Separating them is the work nobody has done.
Dated to April 2025, with no claim on the phrase#
The argument has a date. In Box of Amazing on 20 April 2025, in the essay that first set out the SuperSkills idea, Rahim Hirji wrote that "the edge doesn't come from having knowledge. It comes from turning it into action, at the right time, with the right judgement and wisdom", and that future knowledge workers would be valued for their judgement about which analysis to cultivate and which possibilities to prune.
The phrase "judgement premium" carries no claim of first use. The archive holds no dated first publication of it, so it is used here as a description of a measurable idea and not as a coinage. The estate credits a term to Rahim Hirji only where a dated publication exists, and this is not one of those.
What would settle it#
A realised-wage series for judgement-bearing tasks, rather than for advertisements. A cohort followed through a period of adoption, so the compression finding and the professionalisation finding can be tested on the same people. And a measure of judgement that does not collapse into seniority, since the entry-level result above is about employers asking juniors for senior skills, which may describe a rising bar rather than a rising price. Until then the honest description of the premium is a well-evidenced direction with no magnitude.
Key sources
- PwC (2026). Global AI Jobs Barometer 2026. 15 June 2026. Graded entry.
- Autor, D. and Thompson, N. (2025). Expertise. NBER Working Paper 33941, June 2025; Journal of the European Economic Association, 23(4), 1203-1271, DOI 10.1093/jeea/jvaf023. Read at source 15 September 2026. Graded entry.
- Autor, D. (2024). Applying AI to Rebuild Middle Class Jobs. NBER Working Paper 32140, February 2024, and still a working paper. Its acknowledgements record that it was drafted for NOEMA Magazine, so it is an essay circulated through the series rather than a study written for it. Read at source 15 September 2026. Graded entry.
Related SuperSkills research#
The capability itself is defined at what is judgement. The full evidence on which tasks appreciate sits at what becomes more valuable as AI gets cheaper, and the career reading at staying valuable in the age of AI. On the entry-level side of the same data, the missing rungs and synthetic seniority. On what erodes when the judgement is delegated, capability debt.
About this research#
Written by Rahim Hirji, author of SuperSkills (Kogan Page, 2026), keynote speaker on AI and human capability, and founder of The SuperSkills Intelligence Company. The phrase "judgement premium" is used descriptively here and is not claimed as a coinage; the underlying argument is dated to 20 April 2025 in his own newsletter. Findings are attributed to the studies that produced them and kept separate from the interpretation.
Explainer · SS-2026-208 · Graded against the published rubric
Hirji, R. (2026). What is the judgement premium?. The SuperSkills evidence base, SS-2026-208. https://thesuperskills.com/research/what-is-the-judgement-premium. Last reviewed 9 September 2026.
An evidence review by Rahim Hirji, not peer-reviewed research. For a material claim, cite the underlying study as well; every study here carries its own permanent link.
How citations and IDs work