The fourth Wharton study of enterprise AI finds that fewer than a third of managers report a strong return from AI adoption.
The Adaptive Advantage: Who’s Winning with AI and How is the fourth annual study of enterprise AI adoption by Wharton Human-AI Research and GBK Collective.
The analysis is based on 1,400 responses from employees (409 entry and mid-level and 1,007 managers and above) in large US private-sector organisations. The value to a European public relations context lies in what it tells us about companies that have successfully invested in AI and notably, where they’ve made mistakes.
AI is now part of everyday work in large US organisations. Almost two-thirds (62%) of managers and above use it daily. Three-quarters (73%) say it has significantly changed their workflows. Four in five (80%) report a positive return on investment. But only 28% report a strong positive return.
The researchers score organisations on whether they prove value, transform workflows and align the organisation behind an AI strategyLess than one in ten (7%) clear all three of these bars.
Anyone who’s managed AI adoption in an agency or corporate communications context will recognise this set of numbers and the barriers to adoptions. The report suggests that the wrong things get measured (efficiency versus effectiveness), nobody owns the change management process (implementation stalls as a result) and managers and employees have different views of success.
AI transformation cannot be measured by time saved alone
Around eight in ten managers (83%) expect AI to transform their industry within five years. Yet the most common way to measure the success of AI adoption is the time saved on a task.
I have written about this issue before. Yes, track time saved as an efficiency metric, but if an agency or communication function wants to position itself as a growth driver, rather than an operational cost, it needs to report more than efficiency benefits. Organisations measuring at least one growth outcome such as revenue or customer satisfaction, report a strongly positive return at a rate of 35%. Among departments tracking efficiency alone this falls to 13%.
Somebody has to own AI adoption
Two-thirds of organisations now have a Chief AI Officer or equivalent. The role is typically added to an existing manager’s job rather than a new role. More than half (58%) of managers report a strongly positive return where it’s a dedicated post. That drops to 24% when it’s added to an existing role.
The cost of AI has overtaken skills as the leading barrier to stronger adoption. More than a third (34%) cite tools and licences as an issue. Only four in ten are very confident they can track current spending. Six in ten organisations are trying to prove a return against a figure that they aren’t confident in.
This is the live conversation that’s taking place as organisations plan their 2027 budgets.
Autonomy, risk and accountability
Two-thirds (67%) of enterprise teams use AI agents. Around a fifth (17%) let them act without a person approving each step. Managers cite data security (43%) as the main barrier to greater autonomy. Regulation (24%) and distrust of AI’s reasoning (23%) also feature.
The study reports that senior leaders and the managers running the agents disagree about who’s responsible when something goes wrong. Accountability remains unresolved. This is a significant risk issue for our function, which is typically called in to manage the consequences of a failure.
The management reality gap
Almost four in ten managers (39%) report that their organisation is aligned behind an AI strategy. They describe adequate training, a clear vision and meaningful choice over how people use AI. This number drops to a quarter (26%) among entry- and mid-level employees.
Training is the clearest gap. More than three-quarters of managers (78%) say employees receive adequate AI training. Only 56% of employees agree. The mood among junior employees is cautious rather than hostile.
Public relations practice matches the Wharton study on adoption but lags on what follows: agents, ownership and measurement. Forewarned is forearmed. The issues highlighted in the enterprise data are the ones heading our way.
You can download the full report from The Wharton School of the University of Pennsylvania website. Ben Verinder and I cover all of the issues that it raises, and more, on our AI for public relations one-day online course (see below).
Have a good week ahead.
Upcoming AI and management communications training courses
The Wharton findings come down to management: who owns AI, how it’s measured and how teams are brought along. Our one-day online courses are built around those questions.
Communication management and leadership covers the fundamentals of running a function, from strategy and planning to measurement and reporting to the board. AI for public relations, which I teach with Ben Verinder, covers governance, workflow and implementation for agencies and in-house teams.
Upcoming dates below.
9 December - Communication management and leadership one-day online course
16 February - AI for public relations one-day online course (with Ben Verinder)
24 February - Communication management and leadership one-day online course
18 March - Communication management and leadership one-day online course
28 March - AI for public relations one-day online course (with Ben Verinder)
Further reading
This article was originally posted on my Substack. The Wadds Inc. newsletter is read by almost 6,000 communications and public relations practitioners. We take a slower, critical perspective on the research, evidence and developments shaping the field. We put evidence before opinion, and we say how confident we are in what we publish.
