2026 Decision Intelligence Benchmark — Special AI Report
How social media leaders are adopting AI — or not — to achieve their business objectives
Table of ContentsSocial media leaders have higher AI adoption rates than most of their functional peers. Yet the function’s relationship with AI is defined by a telling gap between what leaders expect and what they feel. Managers are closing the adoption gap on their own, particularly on enterprise AI, where their daily use outpaces directors by 22 points. Directors, closer to the rollout decisions, see more headroom between where teams are and where AI could take them.
The function’s AI proficiency score — second highest of any function surveyed — is paired with above-average dependence concentrated at the manager layer, and conservative automation expectations. Social media teams are using AI actively — but mostly at the execution layer, where speed and volume are the primary gains.
The result: AI is embedded in workflows but not yet embedded in strategy. Members describe AI as a tool for execution — speeding up content and cutting manual steps — rather than a tool for judgment, planning, or creative direction.
The opportunity for social media leaders is not to increase AI usage broadly but to raise the sophistication of how AI is used: from a drafting aid to a strategic input, from a test-and-learn tool to a repeatable workflow.
To better understand how leaders think about different kinds of AI, we grouped tools into three categories used throughout this report:
Social media leaders have strong expectations for AI adoption across all three categories. For public GenAI, actual daily use already nearly meets expectations. Managers are consistently outpacing directors in actual adoption of enterprise and company-owned tools.
Social media leaders hold clear and relatively consistent expectations for AI use across all three types. Public generative AI carries the highest expectation: 63% of leaders expect their teams to use it daily. Enterprise AI follows at 55%, and company-owned AI at 49%.
The seniority cut tells a more precise story. Managers lead expectations for public GenAI at 68% expecting daily use, compared with 58% of directors. On company-owned AI, the pattern reverses: directors expect daily use at a higher rate (55%) than managers (42%).
Public GenAI is where social media teams are using AI most, and the daily adoption rate of 55% trails the 63% expectation by 8 points — the narrowest expectation-to-actual gap of any AI type in this function. Enterprise AI sits at 38% daily use overall, a meaningful shortfall against the 55% expectation. Company-owned AI shows 27% daily use against a 49% expectation, a pattern that mirrors the broader cross-functional trend of internally governed tools trailing in usage.
For enterprise AI, managers report 40% daily use compared with 18% for directors — meaning managers are integrating enterprise AI into their day-to-day work at more than double the rate of directors.
Social media’s expectation-to-actual gap is narrower than average for public GenAI — a distinction in a benchmark where public tool adoption typically trails intent. Enterprise AI shows a 17-point gap between expected (55%) and actual daily use (38%). Company-owned AI shows the largest gap at 22 points, consistent with the pattern across all functions where internally governed tools are built before they are used.
The manager-director divergence on enterprise AI warrants particular attention. Directors should consider gaining experience with AI equal to managers to effectively drive adoption, set targets, and identify workflow efficiencies.
Social media ranks 4th of 11 functions on actual AI adoption — and public GenAI use that nearly meets expectations is a distinction. The manager-director divide on enterprise AI may help identify opportunities for further growth. Members describe primarily using AI for speed and execution — drafting captions, pulling data — rather than informing strategy. This may lead to higher-leverage AI use cases going unexplored.
Social media’s AI dependence is greater than most functions, with 33% of leaders predicting moderate or major disruption if AI disappeared. Managers feel that dependence at nearly double the rate of directors — a reflection of where AI has actually taken root.
Social media sits above the cross-functional average on AI dependence. Thirty-three percent of social media leaders report their function would face moderate or major disruption if AI disappeared, placing the function 4th of 11 across Assemble’s benchmark, reflecting where AI has genuinely taken root in the work.
Managers predict moderate or major disruption at 44%, compared with 23% of directors. Managers are the practitioners who are daily-using enterprise AI tools at higher rates — and their higher dependence reflects that usage. Directors, with lower actual enterprise AI adoption, have correspondingly lower perceived disruption.
Social media leaders’ automation outlook is among the most conservative in the benchmark. Sixty-nine percent expect AI to automate 20% or less of the function’s work in the next 24 months. Only 3% expect more than 40% of work to be automated in that timeframe.
Members’ framing of what AI can do in social media is grounded in task-level efficiency: content creation, posting logistics, analytics reporting. What it cannot do — platform judgment, audience intuition, strategic planning, crisis response — is still the larger share of the function’s real work.
Social media’s 4th-place dependence ranking is higher than the function’s own leaders might expect given how often members describe AI as a “nice to have” rather than a structural tool. But zooming in to managers at the execution layer, AI has become load-bearing. The opportunity is to increasingly extend its reach from tactical content production into the analytical and strategic work.
Social media teams rank high among all functions on AI proficiency, with competent-or-higher ratings across all three AI types. Advanced proficiency concentrates in public GenAI, and directors outpace managers on that type by a notable margin. Company-owned AI has the largest beginner concentration.
Social media teams show a generally competent profile across all three AI types, with competent-or-above ratings reaching 79% for public GenAI, 71% for enterprise AI, and 56% for company-owned AI. The function ranks second among all functions surveyed on overall proficiency — a position that reflects extended exposure to public AI tools and relatively rapid enterprise AI integration through publishing and analytics platforms.
Advanced proficiency is most concentrated in public GenAI, where 24% reach advanced or expert levels. Enterprise AI advanced proficiency sits at 16%, and company-owned AI at 14%. Public GenAI has the longest runway of personal use, self-directed learning, and immediate feedback. Enterprise and company-owned tools require more institutional scaffolding before teams develop the fluency to move beyond competent into advanced use.
The most significant role-level finding in the proficiency data is the director-manager gap on public GenAI. Directors rate their teams at 36% advanced proficiency for public GenAI; managers rate their teams at 21% advanced. The explanation is likely contextual: Directors are more likely to be working with AI in strategic, editorial, and analytical capacities, where advanced prompt construction and output refinement are required. Managers may be high-frequency users for more basic use cases.
On company-owned AI, the largest proficiency gap appears at the manager level: 55% of managers report their teams are beginners or have no experience with company-owned AI, compared with 35% of directors. Given that managers are the highest daily users of enterprise AI, this concentration of beginner-level skill on company-owned tools suggests those tools have not been sufficiently rolled out, trained on, or contextualized at the practitioner layer.
Proficiency and usage are not the same measure, and social media’s data illustrates the gap clearly. Managers are daily users of enterprise AI but beginners on company-owned tools — which means they are driving high adoption without building the deeper fluency. Members describe teams that are in “test-and-learn mode” across AI tools, comfortable with the basics but not yet redesigning workflows around them.
Within the function, trust improves sharply as AI moves closer to organizational governance. Company-owned AI is most trusted, and directors show a steeper public AI skepticism than managers.
The trust gradient in social media runs in a predictable direction: company-owned AI is most trusted, enterprise AI follows, and public GenAI earns the lowest trust. The cross-functional rankings tell a more nuanced story.
On company-owned AI, social media ranks 4th of 11 at 56% trust, above the 49% benchmark average — and 27% of leaders express significant trust, the highest significant-trust figure for any AI type in this function. On enterprise AI, social media ranks 6th of 11 at 52%, essentially at the 50% cross-functional average. On public GenAI, social media ranks 8th of 11 at just 25% trust, 4 points below the 29% average — one of the more skeptical functions on public AI in this benchmark.
Directors are the most skeptical: 59% of directors express moderate or significant mistrust of public GenAI, compared with 39% of managers. At the director level, public GenAI trust falls to just 11% — the lowest of any role cut in this function. Social media leaders in director roles are often responsible for brand safety, compliance oversight, and brand voice and positioning — contexts in which the risks of public AI are more directly in their line of sight. Managers, more likely to be using public tools for production tasks, show more tolerance.
Company-owned AI also shows the sharpest polarization of any function: 56% trust and 13% mistrust coexist, with only 32% neutral. Most functions show more neutral sentiment on company-owned AI; social media leaders have stronger opinions in both directions.
Data privacy or security concerns are the leading driver of public GenAI mistrust at 86%, and policy misalignment follows at 77% — both institutional concerns tied directly to how social media teams handle brand assets, unreleased campaigns, and audience data. Inaccurate or hallucinated outputs also registers at 77% for public GenAI.
For enterprise AI, institutional concerns ease considerably: privacy drops from 86% to 32%, policy misalignment from 77% to 27%. Company-owned AI shows the steepest reductions of all: privacy concern falls to 9%, policy misalignment to 6%. Yet there are some notable areas of mistrust.
Trust and mistrust in social media are not primarily about the quality of AI outputs. They are about what data goes into which AI environment and whether that environment is governed by policies they can follow. Social media leaders who want to raise AI adoption should read its differing opinions as a case for governance design: clear rules about what enters public tools and visible investment in training and adoption for enterprise or company-owned environments.
Social media ranks above the cross-functional average on AI proficiency, adoption, and dependence. Meanwhile, automation predictions sit below average.
Social media’s position in the benchmark is defined by two coexisting signals. The function is performing well on usage and proficiency relative to peers: fourth on adoption, second on proficiency, above average on dependence. Automation expectations sit below the benchmark average, which reflects the function’s honest read of where AI is and isn’t yet redesigning the work.
Across peer functions, those with high dependence tend to be functions where AI has entered data-critical or decision-critical workflows: Employee experience leads at 63% moderate or major disruption, followed by talent marketing at 38% and L&D at 34%. Social media’s 33% places it 4th, just above the cross-functional average of 28%. The function’s above-average dependence is concentrated at the manager layer, where daily AI use has made the tools ingrained in the work.
Social media’s benchmark position describes a function that is ahead on capability and active on usage but has not yet crossed the threshold where AI is integrated into the decisions and workflows that define how the function operates strategically. Proficiency is a genuine advantage — second-best in the benchmark is a meaningful foundation. The question is whether social media leaders will use that proficiency to extend AI deeper into the function’s analytical and strategic work, or whether capable teams will remain at the execution layer, using AI to do existing tasks faster without fundamentally changing how social media strategy is developed and evaluated.