2026 Decision Intelligence Benchmark — Special AI Report

The State of AI in DEI

How DEI leaders are adopting AI — or not — to achieve their business objectives

Table of Contents

This report is interactive. Use the controls to filter all charts by role level. Charts update instantly. Hover over bars to see exact values. Click All Roles to return to the overall view, which is the view reflected in the written narrative.

View by segment
Executive Summary

DEI's Unique Relationship with AI: Ahead on Adoption, Behind on Advanced Skill

DEI leaders are among the more active AI adopters in the benchmark: 57% report their teams using public GenAI daily — tied for the highest rate in the benchmark and 9 points above the 48% cross-functional average, while 50% report their teams using enterprise AI daily — 13 points above the 37% benchmark average. This level of daily adoption outpaces most peer functions, including L&D and CSR.

What the adoption numbers do not surface is the proficiency gap sitting beneath them. DEI ranks 10th of 11 on enterprise AI proficiency, with only 50% of leaders rating their teams competent or above — 10 points below the 60% cross-functional average, and no advanced users on enterprise or company-owned AI. Teams are using enterprise AI at the highest rate in the benchmark while developing advanced capability in it at one of the lowest rates. That combination — high use, low advanced proficiency — is the central challenge ahead in DEI's AI story.

The other defining characteristic of DEI's AI position is the nature of its skepticism. Data privacy is the top mistrust driver on public GenAI at 82%, consistent with the benchmark average. But bias in data or training models ranks 2nd at 77% — the highest bias concern of any function surveyed and 18 points above the 59% cross-functional average. It is also the only function where bias outranks inaccuracy (73%) as the second-ranked concern. DEI leaders understand specifically how training data shapes outputs, and they are responsible for the organizational outcomes that biased AI could distort.

What this means for leaders
Different Perspectives on Different Types of AI

For the purposes of this report, and to better understand how leaders think about different kinds of AI, we grouped tools into three categories:

Public Generative AI
Broadly available tools such as ChatGPT, Claude, Gemini, and similar public-facing generative AI platforms.
Enterprise Platform AI
AI features embedded in vendor-supplied systems, such as CRM tools, productivity suites, or other enterprise software.
Company-Owned AI
Proprietary, internally governed, or organization-controlled AI tools built, licensed, or configured specifically for the company or its teams.
Part 1: AI Adoption

DEI Leaders Set High AI Adoption Targets — and Nearly Hit Them

DEI leaders have set some of the most ambitious AI adoption expectations in the benchmark, particularly for public and enterprise tools. The data also shows that actual daily use is keeping pace better than in most peer functions — a sign that individual leaders are driving informal adoption even without well-developed programs.

Highlights from the data
  • Daily use expectations are high across all three AI types: 67% expect daily public GenAI use, 65% expect daily enterprise AI use, and 47% expect daily company-owned AI use — the first two are notably higher than the cross-functional averages.
  • DEI teams' actual usage of AI is closest to expectations: Most functions in our benchmark show gaps between expected AI usage and actual usage. But in DEI, those gaps are smallest: A 10-point gap on public GenAI (2nd smallest of 11 functions) and a 15-point gap on enterprise AI (smallest of 11 functions). Company-owned AI shows a 20-point gap, in line with the benchmark average.

Expectations of AI Use

DEI leaders set above-average expectations for public and enterprise AI, while company-owned AI expectations sit slightly below the benchmark. 67% expect their teams to use public generative AI daily, compared to the cross-functional average of 60% — placing DEI 3rd among 11 functions on this measure. Enterprise AI expectations are nearly as high at 65%, above the 59% benchmark average. Company-owned AI expectations are lower at 47%, slightly below the 50% cross-functional figure.

Breaking down by level, directors were the largest sample in this group and their expectations for public GenAI daily use are the highest within the function, suggesting an environment of practitioner-level adoption.

Expected Daily Use by AI Type
Q: What is the frequency of use you expect from your team for each AI type?
(N=22) · "N/A" responses excluded; share normalized to valid responses per AI type.
Source: Assemble’s Decision Intelligence Benchmark — Q1 2026

Actual AI Use

Actual daily use in DEI is above the cross-functional average for both public GenAI (57% vs. 48% overall) and enterprise AI (50% vs. 37% overall), making DEI one of the stronger-performing functions on realized adoption for these two tool types. Company-owned AI actual daily use sits at 27%, exactly at the benchmark average. AI use in this function is likely driven by individuals using readily available tools for writing, research, and analysis.

Members describe AI as a frequent subject of strategic discussion, but governance and formal enablement have not caught up.

Actual Daily Use by AI Type
Q: What is the actual frequency of use from your team for each AI type?
(N=22) · "N/A" responses excluded; share normalized to valid responses per AI type.
Source: Assemble’s Decision Intelligence Benchmark — Q1 2026

The Gap

The expectation-to-actual gaps in DEI are smaller than in most peer functions. The public GenAI gap is 10 points (67% expected, 57% actual), and the enterprise AI gap is 15 points (65% expected, 50% actual), while the company-owned AI gap is larger at 20 points (47% expected, 27% actual). For comparison, the cross-functional average gaps are 12, 22, and 23 points respectively. Part of this is a product of alignment, but it also reflects something fundamental about DEI teams: They tend to be small (sometimes just teams of one), making alignment easier or even trivial.

The Gap: Expected vs. Actual Daily AI Use
Percentage-point gap between expected and actual daily use, by AI type
(N=22)
Source: Assemble’s Decision Intelligence Benchmark — Q1 2026
Our Take

DEI leaders are clear about what AI should do for their teams. The tighter-than-average expectation gaps on public and enterprise AI point to individual practitioners driving adoption through personal use rather than waiting for organizational guidance. DEI leaders say they are rarely included in enterprise AI decisions from the outset — often brought into problem-solve after tools are already deployed, not to shape them from the beginning. The next level of AI integration requires identification of the best opportunities for enhancing the work of DEI using technology.

Three actions for leaders
Part 2: AI Dependence & Automation Outlook

DEI Leaders Aren't Dependent on AI (Yet) and See Limited Automation Potential

DEI leaders are not heavily reliant on AI, and they do not anticipate that large portions of their work will be automated in the next two years. This response reflects both the human-centered nature of most DEI work and a realistic read of the function's modest process optimization goals.

Highlights from the data
  • Low disruption dependence: 77% of DEI leaders say work would be "business as usual" or "slight disruption" if AI disappeared tomorrow — above the 73% cross-functional average in this low-disruption band.
  • Conservative automation outlook: 77% believe fewer than 20% of their function's work could be automated by AI in 24 months, the second-highest concentration in this lower range of any function surveyed.

Dependence on AI

77% of DEI leaders predict slight disruption or no disruption if AI became unavailable — a conservative read that places DEI in the lower half of the benchmark on AI dependence. Only 18% predict moderate disruption and 5% major disruption.

At the role level, directors' disruption expectations skew even lower. Among the small senior leader cohort, there is slightly more variance of opinions.

Disruption if AI Disappeared Tomorrow
Q: If AI were suddenly unavailable tomorrow, how disrupted would your function be?
(N=22)
Source: Assemble’s Decision Intelligence Benchmark — Q1 2026

AI Automation Outlook

DEI leaders are pessimistic about future automation potential for AI. This group had the second-highest share among the 11 functions surveyed to believe that fewer than 20% of their function's work could be automated by AI over the next 24 months, at 77%. The cross-functional average has 63% in the under-20% band, and 10% seeing 40% or more automatable — so DEI is notably more conservative on both ends of the scale. Only 5% see 41–60% automation potential.

These results portray the relational nature of DEI work: relationship-building, facilitation, culture change, and navigating organizational politics are inherently human activities. Where AI automation potential is more credible — data analysis, reporting, content creation, employee listening synthesis — the function has expressed some interest but could probably use a more refined vision.

Percent of DEI Work Automatable by AI in the Next 24 Months
Q: Approximately what percentage of your function's work could be automated by AI over the next 24 months?
(N=22) · "N/A" responses excluded; distribution normalized to valid responses.
Source: Assemble’s Decision Intelligence Benchmark — Q1 2026
Our Take

The low disruption and low automation outlook in DEI reflects a sentiment that the most important parts of their work cannot be delegated to an algorithm. While usage has been strong, it is not transformative and leaders are perhaps not seeking technology-driven transformation. DEI leaders are also involved or concerned with AI adoption across the enterprise, an opportunity to gain more exposure on the capabilities of emerging technology.

Three actions for leaders
Part 3: Team Proficiency with AI

DEI Teams Are Competent on AI, Advanced Proficiency Is Rare

DEI leaders rate their teams as broadly competent in using public generative AI and company-owned AI, but expert-level proficiency across any of the three tool types is very low.

Highlights from the data
  • Competent is the ceiling on public and company-owned AI; enterprise AI lags: 71% of DEI leaders rate their teams competent or above on public GenAI (6th of 11 functions, at the cross-functional average of 70%) and 67% on company-owned AI (3rd of 11, 10 points above average) — but only 50% reach competent or above on enterprise AI, ranking 10th of 11 and 10 points below the 60% benchmark average.
  • Advanced proficiency is nearly absent across all three AI types: DEI records 6% advanced on public GenAI and 0% on both enterprise and company-owned AI — below the cross-functional averages of 16%, 13%, and 14% respectively, and among the lowest advanced-proficiency scores in the benchmark on enterprise and company-owned tools.

Proficiency Across AI Types

DEI's proficiency profile is uneven across the three AI types. On public GenAI, 71% of DEI leaders rate their teams competent or above — right at the 70% cross-functional average, ranking 6th of 11 functions. On company-owned AI, 67% rate teams as competent or above, ranking 3rd of 11 and 10 points above the 57% benchmark average — a genuine strength that likely reflects the smaller, more deliberate deployments of internally governed tools.

Enterprise AI is the outlier in the wrong direction: only 50% of DEI leaders rate their teams competent or above on enterprise tools, ranking 10th of 11 functions surveyed and 10 points below the 60% cross-functional average. That gap is particularly notable given that enterprise AI records the second-highest daily use rate in DEI at 50% — teams are using these tools more than their proficiency levels would suggest.

Advanced proficiency is nearly absent across all three types. DEI records 6% advanced on public GenAI, below the 16% cross-functional average, and 0% advanced on both enterprise and company-owned AI, against benchmark averages of 13% and 14% respectively.

Team Proficiency by AI Type
Q: How would you rate your team’s overall proficiency in using each AI type?
Beginner / No experience
Competent
Advanced
Expert
(N=22) · "N/A" responses excluded; rows normalized to valid responses per AI type.
Source: Assemble’s Decision Intelligence Benchmark — Q1 2026

Proficiency by Role Level

The role-level cut on proficiency shows directors as the level most likely to rate their teams as competent on public GenAI — consistent with directors being the most active users and the most direct managers of team AI behavior. What is consistent across levels is the near absence of advanced proficiency ratings. This appears to be a ceiling that exists at every level of the function.

Our Take

Competent is a good place to be, but it is also the plateau for most DEI practitioners. The absence of advanced proficiency on enterprise and company-owned AI — tools that organizations are investing heavily in — is the most actionable finding in this section. DEI has an opportunity to improve the ability of existing, competent users, to realize greater gains from new technology.

Three actions for leaders
Part 4: AI Trust & Mistrust

DEI Trusts Enterprise AI Most — and Mistrust Is Driven by the Function's Mission

DEI leaders show a trust gradient that moves in the same direction as the broader benchmark — higher trust in internal tools, more skepticism toward public GenAI — but the reasons for their skepticism about public GenAI are meaningfully different. Bias is the dominant concern, not privacy.

Highlights from the data
  • Company-owned AI earns the strongest trust at 68%: Enterprise AI also performs well where 55% of DEI leaders express moderate or significant trust in enterprise AI — slightly above the 51% cross-functional average.
  • Bias is DEI's most distinctive mistrust driver: Bias in data or training models is cited by 77% of DEI leaders as a driver of public GenAI mistrust — 18 points above the cross-functional average of 59% and the highest of any function. It is also the only function where bias outranks inaccuracy (73%) as the second-ranked concern on public GenAI, behind data privacy at 82%.

Trust Across AI Types

DEI leaders express moderate or significant trust in enterprise AI at 55%, slightly above the cross-functional average of 51%. Company-owned AI earns the highest trust share at 68% (moderate or significant), well above the 50% benchmark figure — a strong result for a community that often does not own the AI tools they use.

Public GenAI sits at 27% moderate or significant trust, ranking 6th of 11 and just below the 28% benchmark average.

Trust by AI Type
Q: For each type of AI, what is your level of trust with it?
Significant mistrust
Moderate mistrust
Neutral
Moderate trust
Significant trust
(N=22) · "N/A" responses excluded; rows normalized to valid responses per AI type.
Source: Assemble’s Decision Intelligence Benchmark — Q1 2026

Drivers of Mistrust

The mistrust driver profile in DEI highlights the function's mission. Bias in data or training models is cited by 77% of DEI leaders as a driver of public GenAI mistrust — a figure that is 18 points above the cross-functional average of 59% and the highest of any function. Inaccurate outputs comes in at 73% and data privacy at 82% (consistent with the benchmark pattern), but the elevation of bias above peer functions is distinct to DEI.

The bias concern extends into enterprise AI as well: 50% of DEI leaders cite bias as a driver of enterprise AI mistrust, above the 45% cross-functional figure. For company-owned AI, bias drops to 36%, consistent with greater organizational control over training data and model governance. Members say they understand well how AI systems can replicate and amplify historical patterns of bias, and work to remove that bias and hold organizations accountable when AI tools produce inequitable outcomes.

Drivers of AI Mistrust — Public GenAI
Q: When you have trust concerns with Public GenAI, what are the primary reasons? (select all that apply)
(N=22) · Multi-select; denominator is total respondents. Reflects drivers cited for Public GenAI specifically.
Source: Assemble’s Decision Intelligence Benchmark — Q1 2026
Our Take

Enterprise AI earns genuine confidence — likely because employer-governed tools feel more accountable to the organization's stated values. Public GenAI earns skepticism that is specific and grounded: DEI leaders understand first-hand how training data shapes AI outputs. HR and business leaders can consider addressing trust gaps with bias evaluation frameworks and governance processes that give DEI leaders a formal role in reviewing new tools before they are deployed at scale.

Two actions for leaders
Part 5: How DEI Compares Across Functions

DEI Stands Out on AI Adoption and Trust

Compared with peer functions, DEI holds an above-average position on actual AI adoption and AI trust — both meaningfully ahead of the cross-functional average. Team proficiency lands close to the benchmark on aggregate, but DEI's advanced-tier proficiency remains among the most restricted in the benchmark — a narrower, more specific gap than the overall proficiency picture suggests.

Highlights from the data
  • DEI ranks 2nd of 11 on overall AI adoption: 57% daily public GenAI use vs. 48% cross-functional average; 50% daily enterprise AI use vs. 37% average — placing DEI among the top-performing functions on realized use.
  • Advanced proficiency is a narrower, distinct gap: Zero advanced or expert ratings on enterprise or company-owned AI — the most restricted proficiency ceiling among the 11 functions benchmarked — even though DEI's overall competent-or-above proficiency sits close to the cross-functional average.
  • Bias as mistrust driver is function-specific: DEI is the only function where bias outranks inaccuracy as a public GenAI mistrust driver, at 77% vs. 73% (behind data privacy at 82%) — a distinction that should shape how AI governance programs in DEI are designed.
Where DEI Stands Out on AI
DEI’s distance from the cross-functional average on six AI vectors, aggregated across all three AI types
DEI ahead of average
DEI behind average
The benchmark average is the mean of the other ten functions surveyed: Social Media, Healthcare Social Media, ESG & Sustainability, CSR & Social Impact, Talent Marketing, Learning & Development, Employee Experience, Data Privacy, Data Strategy, and Supply Chain. Each vector is the percent of valid responses on a standard benchmark question, averaged across the three AI types where applicable. Hover any bar for DEI’s score, the benchmark average, and DEI’s rank.
Source: Assemble’s Decision Intelligence Benchmark — Q1 2026 (N=22 DEI; 11 functions, cells n≥5).

Cross-Functional Position

DEI's cross-functional position reflects a function that is using AI more than many peers but has not yet built the depth of capability that would make that usage transformational. On AI adoption — the share of leaders reporting daily use across all three tool types — DEI performs above the benchmark average, driven largely by strong public and enterprise AI use rates. Trust is a genuine strength as well: DEI ranks 4th of 11 on the aggregate trust measure, well above average, driven by strong enterprise and company-owned AI trust. These are real advantages. The function has not turned away from AI or treated it as irrelevant.

Team proficiency is closer to the benchmark than the underlying story might suggest: DEI's aggregate competent-or-above rate sits almost exactly at the cross-functional average. The real gap is narrower and more specific — advanced-tier proficiency, which is nearly absent across all three AI types. That distinction matters for how leaders respond: general AI enablement programs will move the aggregate number, which is already fine, but won't produce the advanced practitioners the function currently lacks. On dependence and automation outlook, DEI sits modestly below the benchmark, consistent with a function that is using AI without having woven it deeply into its operational foundation. And while trust is strong overall, the mistrust that does remain is rooted in concerns — particularly bias — that are harder to resolve through general AI enablement than through targeted, bias-specific governance.

Our Take

DEI's position in the benchmark captures a function in an earlier stage of AI integration than its adoption numbers alone might suggest. Usage is real, but it is individualistic and informally acquired. Trust in enterprise tools is genuine, but the governance infrastructure that would make that trust durable is not yet in place. The bias concern that animates DEI's AI skepticism is not an obstacle to be overcome — it is a form of domain expertise that should be formalized into the function's institutional role. The functions that are furthest along in AI are not the ones where adoption happened fastest; they are the ones where adoption was paired with governance, fluency, and clear accountability. DEI has the adoption. The governance and fluency are the next chapter.

Three actions for leaders
  • Position DEI as the bias review function for enterprise AI. Formally offer to conduct bias reviews for AI tools before they are deployed across the organization. Build a simple framework — what populations are represented in training data, how outputs vary by demographic group — that can be repeated and documented.
  • Benchmark against the strongest-performing functions on proficiency. L&D and employee experience lead the benchmark on certain proficiency dimensions. Review how those functions in your organization are building AI capability and adapt their approaches to DEI's specific workflows and team size.
  • Use AI to make the business case for DEI itself. Leaders describe a persistent challenge in quantifying DEI impact. AI-powered analytics tools can reduce the time cost of building that case substantially. Investing there turns a technology adoption decision into a business case for the function's continued relevance.