PwC's survey of almost 50,000 workers ties AI gains to training access

The board-deck version of this survey is the adoption line. Some 64% of workers said they had used AI at work in the past 12 months, up 10 percentage points on a year earlier ^[2]. Daily use of generative AI rose from 14% to 22% ^[3]. Those averages hide where the use sits. On PwC's own shares, front-runners make up 14% of respondents but about a third of all daily generative AI users ^[4][2]. In the engine room, only a small number say they use it daily ^[7].

Peter Brown, PwC's global workforce leader, called the result a "two-speed" workforce ^[12]. Of the engine room, he said: "They're not getting the same access to learning. They're not getting the opportunity to innovate. They are not getting to use AI in a meaningful way." ^[8] Training is the part of that complaint the survey measures. Front-runners report access to learning and development resources at roughly twice the engine room's rate ^[3].

A skeptic would say the causation runs the other way: workers whose skills are already in demand get the training and the better projects, and the survey then names them front-runners. The record gives the skeptic real ground. PwC's categories rest partly on respondents' own reported experience with AI and on how in demand they believe their skills are ^[13]. The survey does not establish the reasons behind what workers experienced ^[13]. A group defined partly by reporting strong AI benefits will report strong AI benefits ^[4][13]. The published findings compare learning access across groups but do not report tool access by group. They cannot show that handing out tools without training produced the split. Brown's "two-speed" label also compresses a model that has four groups into two ^[12][9].

In my view, the defensible reading is narrower than two tiers. In this data, learning access and reported AI benefit go together, and which one drives the other is unproven.

The spending decision does not have to wait for that question to be settled. Whatever put people in the engine room, it holds 56% of respondents, four times the front-runner group, and PwC describes it as the core of day-to-day delivery ^[6][1]. "The danger is that you actually could render a big chunk of your workforce largely irrelevant in the world of work," Brown said ^[14]. He said widening access is not an either-or choice between backing front-runners and investing in everyone else ^[18]. The results show, he said, that "there's an enormous amount of value there that can be tapped" if more of the workforce leans in ^[19].

The source of the prescription matters. PwC employs over 360,000 people and pitches itself as "client zero" for AI transformation while advising companies on the same shift ^[15]. Its own changes include cutting the number of offices its entry-level US consultants can join, to improve community and learning opportunities ^[17].

The trade-off this quarter is between licence spend and learning spend. Next quarter's consequence lands on the 56%: a training budget that follows the front-runners adds more to the group already reporting twice the learning access ^[3]. On the rollout itself, Brown said companies should be transparent about why they are using AI and what they hope to achieve ^[20]. "Workers aren't expecting leaders to, I think, sugarcoat everything," he said. "They just want to understand what's going on." ^[21]