What I'm thinking and reading.

Wharton and Accenture launched the Wharton–Accenture Skills Index (WAsX), a recurring, empirical benchmark of which skills employers actually reward, which they don't, and how fast that is shifting. It is the clearest evidence I have seen that the labor market is moving from roles to skills.

Three findings I keep coming back to

1. The signaling gap. Workers compete on "safe" signals that no longer differentiate them. Oversupplied: leadership (+8.9M), communication (+4.7M), problem-solving (+2.5M), project management (+2.0M). Undersupplied: public speaking (-1.4M), digital marketing (-1.1M), public relations (-0.7M), web development (-0.7M).
2. Skills have price tags — and value is role-specific, not universal. Signaling "strategic analysis" raises a sales representative's pay and lowers a validation engineer's. No skill is always high value; each role is its own micro-economy.
3. AI redistributes value instead of erasing it. Gen AI correlates with lower demand for routine, structured cognitive work and higher demand for judgment, coordination, compliance and domain expertise.

My take: this is the data layer under everything I do. Skill mapping needs a price tag per role, not a company-wide competency dictionary, or you spend reskilling budget on skills that are already abundant. Job redesign follows the same logic: sort tasks by whether they need judgment, coordination and domain expertise, then pay for what the redesigned role actually requires.

Read the report

Stanford Digital Economy Lab's TaskNet (Brynjolfsson, Yang and team) rebuilds job task lists from workers' own accounts, not from formal job descriptions.

How it's built

1. Discover: an AI interviewer helps workers remember and list their tasks (132 interviews).
2. Aggregate: reports are grouped by goal and action, not wording (2,726 task reports).
3. Validate: job analysts check each statement and 318 workers rate it. Tasks fewer than 25% find relevant are dropped.

What it found: compared with O*NET, the official US job database, TaskNet added 79 new tasks. 30 official tasks were missing from what workers described. In the authors' words, "every occupation includes work that keeps it running but seldom appears in formal descriptions of the job."

Useful ratings for AI planning: besides importance and frequency, TaskNet asks whether the standard for doing a task well can be stated in advance (articulability) and whether a coworker could check it (verifiability). Low scores mean tacit know-how, the hardest work to hand to AI.

My take Before redesigning roles for AI, ask the people doing the work. Worker interviews show the hidden tasks and the tacit know-how your skill map would miss, and they bring employees into the change from day one.

TechWolf's Work Intelligence shows a pattern I use with clients: start from tasks, not job titles.

How it works, in four steps

1. Data sources: job postings from the labor market, plus internal data (job architecture, HR systems, workflow tools).
2. Task extraction & normalization: AI breaks each job into tasks and treats tasks with different wording as the same task (e.g. "prepare quarterly statements" = "compile quarter-end reports"). This makes roles comparable across teams and companies.
3. Human vs. AI work: each task is sorted as Human, Augmentable or Automatable, using six lenses: cognitive complexity, uncertainty/risk, domain expertise, empathy, physical work and ethical sensitivity (based on Stanford's Human Agency Scale).
4. Work intelligence: see where the impact is biggest, target upskilling and move people into roles where their human strengths count most.

My take This is the data layer that AI skill mapping and role redesign need. Tools like this give the map, and change management makes people actually move. Pair the task heatmap with sponsor activation and role-based fluency paths, or the insight stays on a slide.

Adoption stalls when tools are bolted onto old workflows. Leaders need to redesign the work itself, measure habit formation, and give champions a feedback loop to the executive table.