When your board reads two AI jobs reports and hears two different messages
A People Analytics Lead walks into a board prep meeting with three tabs open: a BCG piece on AI and the workforce, a PwC CEO survey, and a slide someone on the leadership team pulled from a McKinsey report six months ago. Each uses a different number, a different time horizon, and a different definition of "exposed." The board doesn't want a literature review — it wants one sentence that tells them what to expect and what to do about it.
This is the moment where credibility is won or lost. Citing two macro reports that appear to contradict each other, without explaining why, reads as either sloppy or alarmist. The fix isn't picking the scarier number. It's understanding what each publisher actually measured, so you can say — accurately — where they agree, where they don't, and why that's expected rather than a red flag.
This article walks through what PwC's CEO-level research measures, what we can and can't verify from BCG's AI workforce publications at the time of writing, and a framework for reconciling reports like these into one board-ready narrative rather than a stack of citations.
What PwC's CEO surveys actually measure
PwC's Global CEO Survey is a survey of executive intentions and reported outcomes, not a task-level exposure model. In its 27th edition (2024), PwC found that 25% of CEOs expected a headcount reduction of 5% or more attributable to generative AI, while 39% expected a headcount increase of 5% or more over the same horizon. A year later, in the 28th edition (2025), the picture shifted from expectation to reported experience: 13% of CEOs said generative AI had already reduced headcount (rising to 16% in the sectors most affected), while 17% reported an increase.
Two things matter about reading these numbers correctly. First, they're self-reported executive sentiment, aggregated across industries and geographies — not a measurement of any single company's task exposure. Second, the shift between the two surveys (from "expect" to "have observed") is itself informative: it tells you sentiment and reality don't move in lockstep, and that a board conversation grounded only in expectation surveys will age quickly.
There's also a separate PwC publication — commonly referred to in the market as an AI Jobs Barometer — that reportedly tracks AI-related shifts in job postings and wage premiums by occupation. We don't have independently verified figures from that specific publication in our source library, so we're not citing numbers from it here. If your board deck needs figures from it, pull them directly from PwC's current release and confirm the publication date and methodology before quoting.
What the BCG AI jobs report actually focuses on — and where the record gets thin
BCG has published workforce-and-AI research addressing similar territory to PwC and McKinsey: adoption patterns, skills implications, and organizational readiness. What we can say with confidence, based on how BCG typically frames this work, is that it tends to combine survey data from executives and employees with qualitative analysis of skills and workforce strategy, rather than occupation-by-occupation exposure scoring.
What we're not going to do is attach a specific percentage or dollar figure to "the BCG AI jobs report" in this article. We don't have a verified figure from that publication in our source library, and inventing one — even a plausible-sounding one — would be worse for your credibility than citing nothing. If a specific BCG statistic matters to your board narrative, go to BCG's published research directly, note the publication date, and cite it from the source. That's not a knock on BCG's research; it's a discipline we apply to every publisher, including PwC, McKinsey, and Goldman Sachs, when we can't trace a number back to the original release.
This is worth normalizing in your own reporting practice, too. A board deck that says "BCG's research addresses similar themes; we're pulling the exact current figures from their site for this deck" reads as more careful than one that quotes a number nobody on the team can trace to a page.
Why macro reports disagree — and why that's not a problem
Once you separate "survey of intentions" from "task-based estimate," most apparent contradictions between AI jobs reports resolve on their own. Consider three publishers with figures we can verify:
- McKinsey (2023) estimated that 60–70% of current work activity hours could theoretically be automated with existing and emerging technology — a technical-feasibility ceiling, not a forecast of what will happen by a given date.
- Goldman Sachs (2023) estimated the equivalent of 300 million full-time jobs globally are exposed to some degree of automation from generative AI — an economy-wide exposure estimate, not a company-level prediction.
- The World Economic Forum's Future of Jobs Report 2025 projected 170 million jobs created and 92 million displaced by 2030 (22% total churn, net +78 million), alongside a finding that 39% of workers' current skills will be transformed or become outdated in the same period.
None of these numbers "disagree" with PwC's CEO survey figures. They're answering different questions — technical feasibility, economy-wide exposure, net job churn, and executive-reported headcount change are four different measurements that can all be true simultaneously. The mistake is treating any one of them as the answer your board is asking for. Read our summaries of the McKinsey AI workforce report and the Goldman Sachs AI jobs report for the full methodology behind each figure, and see how Anthropic's Economic Index approaches the question from actual AI usage data rather than survey response.
A framework for reconciling reports before they reach the board
Before you cite any macro report in a board deck, sort it along two axes: what it measures (executive sentiment, technical feasibility, actual usage data, or projected job churn) and what it covers (the whole economy, a sector, or a single company). Almost every apparent contradiction between BCG, PwC, McKinsey, Goldman Sachs, and WEF research collapses once you place each figure correctly on that grid.
A useful discipline: for every macro statistic you put in a slide, add one line naming the publisher, the year, and what was actually measured — sentiment, feasibility, or usage. That single habit prevents the most common board-meeting failure mode, which is a leader asking "how did you get this number?" and the presenter not being able to answer precisely.
From macro percentages to your organization's actual exposure map
Macro research — BCG's, PwC's, McKinsey's, or anyone else's — answers "what might happen across the economy." It cannot tell your board which of your company's roles carry the highest exposure, because it wasn't built to look at your role list. That's a structurally different question, and it requires structurally different data: your own job titles, mapped to O*NET's occupational and task framework, scored against a consistent rubric.
WorkforceAnalysis takes that second step. It maps your organization's roles to ONET's occupational database and scores each one across four dimensions — cognitive routine, physical routine, social/judgment, and creative — each on a 0–100 scale, weighted by the task importance data ONET already publishes for that occupation. The output is a department-level heatmap and a Monitor / Review / Redeployment Candidate categorization for each role — an input to your workforce planning conversation, not a verdict on any individual's job. For more on how macro trends connect to task-level analysis, see how AI is changing the future of work.
Bringing it back to your next board meeting
The version of this story that holds up under a CFO's follow-up question is never "BCG says X and PwC says Y, so the truth is somewhere in between." It's "here's what each report actually measured, here's how they fit together, and here's what we found when we ran the same lens against our own roles." Macro research sets the context; your own data closes the loop.
If you're building that board narrative now, our Board Deck & Executive-Summary Template Pack is built specifically to help you cite macro research correctly alongside your own findings. And if you want the next report breakdown — BCG, PwC, or otherwise — delivered straight to your inbox as we verify and publish it, subscribe to our newsletter below. You can also see current plans on our pricing page if you're ready to run your own exposure map ahead of the next board cycle.