PwC

Confidence 0.80 · 3 sources · last confirmed 2026-08-12

Global professional-services firm (one of the Big Four). First entered the wiki via Dan Priest’s interview on The Most Interesting Thing in AI (Nicholas Thompson, June 2026) — content produced by Atlantic Re:think in collaboration with PwC, so treat firm-specific claims here as directionally credible practitioner testimony, not independently audited data (see that source’s Source quality note).

Appears in this wiki via

• 2026-07-22-brown-wef-meet-the-leader-entry-level-jobs-in-an-ai-era — Peter Brown, PwC’s global workforce leader, interviewed for the World Economic Forum’s Meet the Leader podcast (22 Jul 2026) on a WEF–PwC report on AI and entry-level skills. The wiki’s second PwC-sourced dataset after the Priest interview, and the one with the most quotable figures: 3× productivity growth in AI-exposed sectors among organisations that have fundamentally redesigned work; 2× growth in demand for human skills (judgment, problem solving, critical thinking, relationship building); employers asking for 7× more “seniorized” skills at entry level — skills that used to accrete over three to four years; and 81% prioritising applied experience against 34% prioritising degrees. Brown also cites PwC’s AI Jobs Barometer and its 55,000-worker Hopes and Fears survey, and describes PwC’s own entry routes being cut from “17, 18 ways you could join” to about five. Second-hand and unverified here — see that source’s quality note, and the typed contradicts edge to Brynjolfsson’s ADP payroll data.
• 2026-08-01-bbc-ai-decoded-why-isnt-ai-working-for-your-company — cited rather than authored: PwC is the source of the survey that supplies this BBC panel’s headline ROI statistics — ~4,500 CEOs across 95 countries, released at Davos/the World Economic Forum, finding 56% yet to see a return on their generative-AI investment and only one in eight able to claim an actual cost saving or revenue gain. Peter Grant explicitly prefers it to the widely-cited MIT 95% figure on sample-size grounds (153 companies). Not independently checked in this ingest, and PwC is itself a seller in this market. See micro-productivity-trap and enterprise-ai-adoption.
• 2026-06-17-priest-atlantic-pwc-ai-agents-changing-business — Dan Priest (Chief AI Officer) describes PwC’s own AI-transformation practice: a firm-wide GPT that compounds the transformation team’s learnings and distributes them to every consultant; a task-registration “operating system” for managing agents across multiple LLM platforms with per-model accuracy/latency/drift/task-length tracking; a Southwest Airlines case study (50% time/effort reduction in design phase alone, 30–50% benefits from code generation); net hiring growth of ~5,000 people with agent-management skills now tested in interviews; and the hourglass organization model (expanded entry-level intake, compressed-but-empowered middle management, growing leadership layer) as PwC’s emerging prescription for AI-era org design.

Open questions

• The Priest interview’s firm-specific numbers (Southwest Airlines case study, hiring figures, 85–90% agent accuracy) are unverified against any published PwC report. Watch for independent corroboration or a PwC-published report to cite directly.
• Three PwC-sourced datasets now reach the wiki only through interviews (Priest, the BBC panel’s Davos CEO survey, and Brown’s WEF report). None of the underlying reports has been ingested, so sampling frames and definitions — notably “AI-exposed sector” and “seniorized skills” — remain unexamined across all three.
• The CEO-survey figures cited on the BBC panel (56% / one-in-eight, ~4,500 CEOs, 95 countries) are reported second-hand by a guest and have not been traced to the published PwC release. Locating the primary report would let the wiki cite the sampling frame and the definition of “return” directly — both load-bearing for how the statistic is read.