2026 Decision Intelligence Benchmark — Special AI Report
How EX leaders are adopting AI — and where a 50-point company-owned AI gap signals unfinished infrastructure
Table of ContentsEmployee experience leaders set the benchmark's highest expectations for company-owned AI daily use at 70% — and tie for the highest on public GenAI at 88%. The story told by the actual adoption data is more mixed. Public GenAI delivers: 62% of EX leaders report daily use, 1st of 11 functions and 14 points above the 48% cross-functional average. Enterprise AI holds at the benchmark average at 38% daily use, 5th of 11.
But company-owned AI shows a −50-point gap between what leaders expect (70%) and what they report actually using daily (20%) — the largest company-owned AI gap in the benchmark, and more than double the −23-point cross-functional average. EX set the highest company-owned AI expectations of any function and has the proficiency to back them up — ranking 1st of 11 on company-owned AI proficiency at 89% competent or above. The gap appears to reflect a deployment challenge rather than a capability problem.
The benchmark suggests AI has become more deeply embedded in how EX leaders work than in any other function in the benchmark. 63% expect moderate or major disruption if AI disappeared tomorrow — 1st of 11 functions and 37 points above the 26% cross-functional average.
The data indicates AI as consequential to daily workflow, consistent with a function that is increasingly expected to operate with a product orientation, defining where technology belongs in the employee lifecycle and where human judgment must stay. Connecting new AI tools to existing HR technology platforms is the integration challenge EX leaders name most often.
Trust adds a further layer of complexity. Enterprise AI earns only 31% moderate or significant trust among EX leaders — 10th of 11 functions and 20 points below the 51% cross-functional average. That is the function's most notable trust deficit, and it sits in tension with the expectation that enterprise AI should be embedded in the employee lifecycle. Leaders are potentially being asked to deploy AI tools within platforms they do not fully trust, for employee interactions they know require human care.
For the purposes of this report, we grouped AI tools into three categories:
Employee experience leaders set the highest expectations of any function for company-owned AI and tie for the highest on public GenAI. Actual daily use on public tools meets those ambitions better than most functions.
EX leaders hold some of the most ambitious AI adoption expectations in the benchmark. 88% expect their teams to use public generative AI daily — 28 points above the 60% cross-functional average. Company-owned AI expectations are even more distinctive: 70% expect daily use, the highest figure of any function and 20 points above the 50% benchmark average. Enterprise AI expectations, at 62%, are more modest — 7th of 11 functions and just above the 59% average.
The role-level cut on expectations is informative on company-owned AI. Managers report 100% daily use expectations on both public and company-owned AI, while directors sit at 67% for both.
62% of EX leaders report daily public GenAI use — 1st of 11 functions. That is the benchmark's strongest realized adoption on public tools, and it suggests public AI has become part of many leaders' regular working practices. Enterprise AI actual use sits at 38%, in line with the 37% cross-functional average and 5th of 11 functions — a middle-of-the-pack result. Company-owned AI actual daily use is 20%, 9th of 11 functions and 7 points below the 27% cross-functional average.
The role-level actual use data on company-owned AI surfaces the largest role-level difference within employee experience.
The public GenAI expectation-to-actual gap in EX is −26 points (88% expected, 62% actual). That is larger than the −12-point cross-functional average, but it reflects the unusually high starting point of the expectations rather than weak adoption — 62% actual daily use is the benchmark's best result. The enterprise AI gap is −24 points, close to the −22-point average. The company-owned AI gap is −50 points — more than double the −23-point benchmark average and the largest of any function.
EX's public GenAI adoption story is strong — leading the benchmark on realized daily use is a telling indicator. Employee experience set the highest expectations in the benchmark for internal AI use, and the actual use rate tells the story of a function whose high ambitions have not been met, even though usage is ahead or on-par with other functions.
The data suggests the integration challenge directly: connecting new AI tools to an existing HR technology stack is consistently named as the hardest part of deployment. The approval process for new tools adds further friction — a pattern that EX shares with L&D and talent marketing functions.
Employee experience leaders are more dependent on AI than any other function in the benchmark. 63% expect moderate or major disruption if AI disappeared tomorrow — well above the cross-functional average of 26%. The automation outlook is more conservative, with most leaders expecting fewer than 20% of their work to be automated in the next 24 months.
In addition to having the highest AI dependence in the benchmark, only 37% of EX leaders say their work would be business as usual or slightly disrupted if AI disappeared tomorrow — the lowest share of all 11 functions, compared to the 73% cross-functional average. 44% expect moderate disruption and 19% major disruption — both well above the benchmark averages of 22% and 4% respectively.
Despite the high disruption dependence, EX leaders are relatively conservative about long-term automation potential. 69% believe fewer than 20% of their function's work could be automated by AI in the next 24 months — 3rd most conservative of 11 functions and above the 63% cross-functional average. 25% see 21–40% automation potential, and 6% see 41–60%. No survey respondent sees more than 60% of their work becoming automatable.
The combination of high disruption dependence and conservative automation outlook describes a function where AI has become integral to how work gets done today — in productivity, communication, and workflow — without leaders yet believing that AI will replace entire work streams.
EX's disruption dependence is the most distinctive data point in this report. It tells a story that the adoption numbers only partially capture: AI has become genuinely woven into how EX leaders do their work, more so than in any other function surveyed. The automation outlook adds important context. EX leaders are not worried that AI will replace what they do — they are dependent on AI as infrastructure for doing it. That level of dependence warrants a resilience strategy, not just an adoption plan.
Employee experience leaders rate their teams at or near the top of the benchmark on proficiency for both public generative AI and company-owned AI. Enterprise AI proficiency sits near the middle of the pack — the one area where EX's capability does not match its ambitions or its peers.
EX leaders rate their teams highest on public GenAI and company-owned AI proficiency — 89% competent or above on both, ranking 1st of 11 functions on each. Advanced proficiency on both types is 22%, above the benchmark averages of 16% and 14%, making it one of the few functions with expert-level public GenAI proficiency.
Enterprise AI is the exception. 61% of EX leaders rate their teams competent or above on enterprise tools — 5th of 11 functions and at the 60% benchmark average. 38% are at the beginner level, the highest beginner share for any AI type in EX. Advanced enterprise AI proficiency is 15%. This has practical significance given that enterprise AI is the tool type where EX's trust deficit is also most acute.
Directors in EX show the widest proficiency spread across AI types. On public GenAI, 75% of directors rate their teams competent or above, with 25% at advanced. On company-owned AI, 83% of directors reach competent or above, with 33% at advanced — the strongest advanced score in the role-level cut. On enterprise AI, directors sit at 50% competent or above, with 17% advanced — a meaningful gap below their performance on the other two types. This suggests that directors may currently show the strongest proficiency profile in EX for working with public and internal AI but have built less relative capability in the enterprise tools embedded in their HR platforms.
EX's proficiency strengths on public and company-owned AI are the clearest good news in this report. Leading the benchmark on both competent-and-above and advanced proficiency for these two tool types reflects a function where leaders have invested real effort in building team capability — not just access. The enterprise AI gap is worth naming precisely because the rest of the proficiency picture is so strong. Members describe a function where the HR-to-product orientation is shifting — where leaders are expected to understand AI tools deeply enough to define which workflows they should touch. Enterprise AI is the tool type most embedded in those workflows. Closing the proficiency gap there is not just a training decision; it is a strategic positioning decision about how much EX leaders want to shape how enterprise AI is configured for employee experience purposes.
Employee experience leaders show a trust profile that moves in the expected direction — more trust in internally governed AI, less in public tools — but the enterprise AI trust figure is the lowest notable result in the function's data, coming in 20 points below the benchmark average.
31% of EX leaders express moderate or significant trust in enterprise AI — 10th of 11 functions and 20 points below the 51% cross-functional average. Only Healthcare Social Media scores lower than EX on enterprise AI trust at 25%. The figure is particularly notable because enterprise AI is the tool type that EX leaders are most likely to be deploying within HR platforms — in performance management, employee listening, and service delivery — and it is the tool type where EX's use rate (38%) most closely matches the cross-functional norm.
Company-owned AI earns the most trust in EX at 44% moderate or significant — above public GenAI at 31%, and consistent with the general benchmark pattern of higher trust in internally governed tools. However, EX's 44% figure is still below the 50% cross-functional average, ranking 8th of 11. Senior leaders and managers both reach 67% moderate or significant trust, while directors sit at 43%. That pattern — directors trusting company-owned AI less than the levels above and below them — mirrors the director-level disruption exposure identified in Part 2 and may reflect directors' closer visibility into where internal AI tools fall short in day-to-day use.
Mistrust in EX has two notable elevations relative to the cross-functional average. On enterprise AI, 31% of EX leaders cite threat to job security as a driver of mistrust — 12 points above the 19% benchmark average and the highest job security concern on enterprise AI of any function.
Bias in data or training models is also elevated on enterprise AI at 62%, vs. a 45% cross-functional average. On public GenAI, inaccuracy (62%) and data privacy (75%) are the top concerns, both slightly below benchmark averages of 76% and 82% respectively — suggesting EX leaders are not disproportionately worried about public AI.
The job security elevation on enterprise AI deserves attention. Members describe a function where the HR-to-product orientation shift is changing what is expected of leaders and their teams. AI in performance management, talent assessment, and HR service delivery is affecting the work these leaders manage and, in some cases, the roles their teams occupy. The concern is grounded in the real proximity of enterprise AI to the workflows EX leaders govern, not in general AI anxiety.
The job security trust driver in EX and lack of trust in enterprise AI are the most actionable findings in Part 4, in part because the gap is so large relative to the benchmark. Leaders are not worried about enterprise AI in the abstract. They are worried about bias in the tools that make decisions about their employees, and about the job security implications of AI in HR workflows. The response that matches the scale of the concern is not a trust-building communications exercise — it is governance: visible oversight of how enterprise AI tools are configured, tested for bias, and monitored when they affect employee outcomes.
Employee experience holds the benchmark's strongest position on actual AI adoption and team proficiency for public and company-owned AI. Its most significant gaps relative to peers are enterprise AI trust — where it ranks 10th — and company-owned AI deployment, where a 50-point expectation gap stands alone in the benchmark. We score every function on the same six AI vectors, each measured the same way, then plot EX’s distance from the cross-functional average — bars to the right mean EX leads its peers.
Employee experience's cross-functional position reflects a function that has built genuine AI capability and integrated AI deeply into how its leaders work. On the six vectors used to map functional AI position, EX's distance above the cross-functional average is largest on AI Adoption (actual daily use) and Team Proficiency, where it leads the benchmark. AI Dependence — the share expecting moderate or worse disruption — is the highest in the benchmark, though this metric cuts both ways: it reflects real integration and vulnerability simultaneously.
EX sits below the cross-functional average on AI Trust for enterprise tools, and its Automation Outlook (share expecting more than 20% of work to be automatable) is near the benchmark average at 31%. The overall picture is a function at the high end of the benchmark on the dimensions that reflect what leaders have built for themselves, and below average on the dimensions that reflect what organizations have built for them.
EX's benchmark position captures a function that has done the individual-level work of AI adoption and capability building — and is now waiting for organizational infrastructure to catch up. Leading the benchmark on actual public GenAI adoption, on public and company-owned AI proficiency, and on AI dependence tells a coherent story: these leaders have made AI part of how they work. The company-owned AI deployment gap and the enterprise AI trust deficit tell the complementary story: the systems side of AI has not kept pace with the people side. Members describe this gap clearly — integration with existing HR tech is the hardest part, approval processes are slow, and they are being asked to deploy enterprise AI tools in workflows that affect employees without sufficient visibility into how those tools are governed. Closing the distance between EX's adoption and infrastructure is the function's defining AI challenge for 2026.