The labor data just named your appreciating asset. It's the part you can't put in a textbook.

On August 12 the Stanford Digital Economy Lab published a revised version of Canaries in the Coal Mine. The coverage that followed took the half of the finding that fits a headline. That's nearly always the half about what's shrinking.

The other half is the more useful one for anyone running an academic institution. Employment rose among workers in occupations that rely on tacit knowledge. The authors define that as knowledge acquired through practice, mentorship and repeated exposure to real situations.

Read that definition again with a course catalog open. It describes a clinical rotation. It describes a studio critique, a co-op placement, a field practicum and a research apprenticeship. These are the hardest parts of a university to scale. They are also the hardest to justify on cost per credit hour.

The labor data now says they're the parts appreciating fastest.

The evidenceWhat the August revision separates

Erik Brynjolfsson, Bharat Chandar and Ruyu Chen use ADP payroll records running through June 2026. Their analysis sorts occupations by exposure to generative AI, then tracks employment by age.

The number that traveled concerned young workers in highly exposed occupations, where employment sits roughly 19% below its counterfactual. That figure's real, and worth knowing. It's also the less useful of the two.

The finding underneath it is more actionable. Where AI is used to complement workers rather than automate their tasks, employment is flat or rising. For workers aged 22 to 25, the three least-exposed quintiles grew about 10% between November 2022 and June 2026.

The authors offer a mechanism. Generative AI reproduces knowledge already encoded in text and documentation. Knowledge built through supervised practice has proven harder to replicate.

The distinction they draw is between automation and complement. Where AI usage automates human tasks, employment among young workers fell. Where it complements them, employment held or grew.

That's a sorting mechanism, not a verdict on any field. It cuts across disciplines rather than along the usual lines of prestige or salary.

For a university, that sentence is a strategy document wearing the clothes of a footnote. It shows which parts of the offer are commoditizing and which are gaining scarcity value. Few institutions have assembled that view of their own portfolio.

The Stanford authors are careful, and their caution should travel with the number into any board meeting. They describe these as descriptive patterns, not causal estimates. Gaps narrow once education is controlled for. Estimated gaps run larger in the ADP sample than in national survey benchmarks.

The interpretationYour most expensive programs are your most defensible

Experiential learning has been on the wrong side of every efficiency conversation for a decade. Low student-to-faculty ratios. Site coordination. Liability. Equipment. Placement partnerships that take years to build.

Every one of those line items has been defended in a budget meeting by someone who suspected they were losing the argument. Those same characteristics are what make the model hard to replicate. Hard to replicate is now the point.

The inventory is broader than most institutions assume. Nursing and allied health carry supervised clinical hours that no simulation has managed to replace. Engineering runs co-op rotations. Architecture and design are built on the studio critique, which is peer review conducted out loud and in public.

Teacher preparation places candidates in classrooms under a mentor. Laboratory sciences train through bench work you can't simulate from a reading list. Social work, law and business each have practicum or clinic models.

Very few of these programs are marketed as a category. They sit in separate schools, under separate deans, with separate outcome reporting. The portfolio exists. The org chart just never described it that way.

Your students already sense this. Roughly nine in ten enrolled students told Lumina-Gallup they were confident their education will equip them for the job they want. Three-quarters of alumni called their degree critical or important to their career goals. That's a large reservoir of goodwill, and it's evidence you're allowed to cite.

Demand is moving too. Just over four in ten bachelor's students told Gallup in April that AI had influenced their choice of major. Students are hunting for the programs that'll hold their value.

Name those programs, with evidence, and you're answering the exact question families are already asking.

The implicationSpeed is the advantage that is actually available

Here the opportunity turns concrete, and it has less to do with curriculum than with calendars.

Under NACE first-destination standards, outcomes data is collected within six months of graduation. Institutions submit between January and April. Aggregated national results appear in the fall. The compliance target is a 65% knowledge rate.

That cycle serves accreditation well. It wasn't built to answer a question in the middle of a recruitment season.

There's something astronomical about it. The light's real and the measurement's sound, but it left the source a long time ago.

Meanwhile the Stanford Canaries Dashboard updates monthly. Evidence about which occupations are gaining now refreshes faster than most institutions can act on it.

Close that gap and you get to be current. You walk into a recruitment event, a board meeting or a donor conversation with this month's answer. Peers are still citing last year's report.

A faster loop doesn't mean a second survey. It means adding sources that already refresh on their own.

Employer hiring behavior in your placement network. What disciplinary communities say about your programs. What alumni post about their first roles. What AI systems return when a prospective family asks about your school.

None of this takes new academic programs. It takes seeing what you already have clearly, and early enough to talk about it.

The recommendationWhat is winnable this cycle

The evidence supports confidence, as long as you document it. Five moves are open before the next deposit deadline.

1. Inventory your tacit-knowledge assets. List every program with a clinical, studio, field, laboratory or co-op requirement. That list is your appreciating portfolio. Most institutions have never put it on one page.
2. Measure those programs separately. Track placement, wage and progression outcomes for experiential cohorts distinctly from the institutional average. The difference is your proof.
3. Shorten the loop between reports. Add a continuous read on employer and alumni sentiment at program level. The annual survey stays for compliance. This runs alongside it, for decisions.
4. Name the people who already vouch for you. Map the employers, faculty voices and recent alumni shaping the answer to "is this program worth it." Give them something citable to work with.
5. Brief the board with the caveats attached. The Stanford authors call their own findings descriptive. A trustee briefing that carries that qualification holds up better than one that doesn't.

The strongest position here is proof, produced faster than opinion forms. Institutions that can show what their experiential programs deliver will spend the next cycle answering with evidence.

Start with the three programs you would defend first in a room full of skeptics. The data now says you're probably right about them.