Original AISA research · 91 people · August 2026

Barriers to AI Adoption at Work [2026 Data]

AISA asked 91 people about barriers to AI adoption in the workplace. 51.6% say nothing holds them back. See what stops the rest.

By Ozan Dagdeviren··11 min read
insightsAISA researchAI adoptionworkplace AIteam AI skillsbarriers to AI

When AISA asked 91 people starting the AI fluency assessment "What's the biggest thing holding you back from using AI more?", the single largest group (51.6%) said nothing at all. But the other 48.4% named real barriers to AI adoption in the workplace: not knowing where to start (24.2%), no time to learn (12.1%), distrust of AI output (8.8%), and lack of organisational support (3.3%). If you manage a team or lead a transformation programme, those numbers tell you exactly where to focus.

Original AISA research“What's the biggest thing holding you back from using AI more?”
  1. Not sure where to start24.2%22
  2. Don't trust the output8.8%8
  3. No time to learn12.1%11
  4. My org doesn't support it3.3%3
  5. Nothing — I'm all in51.6%47

91 people answered, 6 August to 10 August 2026. Source: AISA research, aisa.to

What Is the Biggest Barrier to AI Adoption at Work?

The biggest single barrier is uncertainty about where to begin. Among the 48.4% of people who named a barrier, "not sure where to start" was the most common answer at 24.2%, roughly one in four people. That makes orientation, not resistance, the primary obstacle.

This matters because it reframes the problem. The conversation about employee resistance to AI often assumes people are pushing back. The data shows something different: most people who aren't fully engaged with AI simply lack a clear first step.

The Four Named Barriers

Each barrier points to a different kind of gap:

  • Not sure where to start (24.2%), an orientation gap. People want to use AI but don't know which tasks to apply it to or which tool to pick up first.
  • No time to learn (12.1%), a capacity gap. The intent is there, but the calendar isn't.
  • Don't trust the output (8.8%), a confidence calibration gap. People have tried AI and found the results unreliable, or they've heard enough cautionary stories to stay cautious.
  • My org doesn't support it (3.3%), a structural gap. Policy, tooling, or culture actively blocks adoption.

What This Means for Transformation Leads

If you're designing an AI adoption programme, these numbers suggest that the largest return comes from solving the orientation problem. Training that starts with "here's what a large language model is" may miss the mark. Training that starts with "here's the first task you should try" is closer to what 24.2% of people are asking for. Our companion post Why Don't People Use AI More? [2026 Data] explores the full picture of AI usage frequency alongside these barriers.

How Many People Say Nothing Is Holding Them Back?

Just over half (51.6%, or 47 out of 91 people) chose "Nothing — I'm all in." That's the largest single response by a wide margin.

This group isn't a problem to solve; it's a resource to deploy. These are the people who can model effective AI use for colleagues, contribute to internal prompt libraries, and help others get past the "not sure where to start" barrier.

What "All In" Doesn't Tell You

Saying nothing holds you back is not the same as using AI well. Someone can be enthusiastic and still lack critical thinking skills when working with AI. Enthusiasm without skill can lead to over-reliance on AI at work, where people accept outputs uncritically. A team AI assessment gives you visibility into whether your "all in" group is actually proficient or just confident.

Turning Enthusiasm Into Leverage

Identify the people in this 51.6% who also demonstrate strong AI fluency. Pair them with colleagues who named a barrier. Peer learning is faster than formal training and costs nothing to start.

Is "Not Sure Where to Start" the Real Barrier?

At 24.2%, it's the most commonly named barrier, and arguably the most actionable one. Unlike distrust or organisational resistance, "not sure where to start" has a straightforward fix: show people a concrete first task.

The challenge is that "where to start" means different things to different roles. A product manager's first useful AI task (drafting user stories, synthesising feedback) looks nothing like a developer's (generating test cases, reviewing code). Generic "intro to AI" sessions don't close this gap.

What Works Instead

Role-specific task decomposition is the most direct answer. Break each role's workflow into tasks, identify which ones benefit from AI, and give people a worked example they can try in under ten minutes. AISA's free guide on using AI at work walks through this approach step by step.

The Training Gap Is Real

The orientation barrier doesn't exist in a vacuum. According to BCG's AI at Work 2025 report (surveying over 10,600 workers across 11 countries), only 36% of employees feel adequately trained in AI use (BCG, June 2025). Similarly, the Microsoft and LinkedIn Work Trend Index found that only 39% of people who use AI at work have had AI training from their company (Microsoft, May 2024). When the majority of people haven't received structured training, it's no surprise that nearly a quarter say they don't know where to begin.

How Much Does Distrust of AI Output Hold People Back?

8.8% of people named distrust of AI output as their biggest barrier. That's a smaller group than you might expect given how much attention hallucination detection and AI accuracy get in the press, but it's a group that matters disproportionately.

Why This Group Matters More Than Its Size Suggests

People who distrust AI output tend to be in roles where accuracy is non-negotiable: legal, finance, compliance, clinical work. Their scepticism is often well-founded. AI models do produce confident-sounding errors. The question isn't whether to override their caution, it's how to channel it.

Building Warranted Trust

Distrust becomes productive when it's paired with verification skills. Teach this group how to check AI output systematically: source triangulation, cross-referencing against primary data, and spotting common failure patterns. AISA's guide on critical thinking with AI covers these techniques. The goal isn't to make sceptics into enthusiasts. It's to give them a reliable process so they can use AI where it helps and reject it where it doesn't.

AISA

Curious about your AI Fluency?

AISA helps you measure, prove and improve your AI skills — free report in a 20–40 minute chat.

What Role Does Time to Learn Play?

12.1% of people said "no time to learn" is their biggest barrier. This is the second most common barrier after orientation, and it's the hardest one for an individual to solve alone, because it's usually a management problem, not a motivation problem.

Time Is a Resource Allocation Decision

When someone says they have no time to learn AI, they're telling you that their current workload doesn't leave room for skill development. That's a signal for managers, not for the individual. If AI adoption is a strategic priority, learning time needs to be protected the same way sprint capacity or professional development budgets are.

Small Bets, Not Big Programmes

You don't need a week-long bootcamp. The data from AISA's research question What AI Skill to Learn First? [2026 Data] shows that people gravitate toward practical, task-level skills. A 30-minute session where someone applies AI to a real task from their backlog is worth more than a half-day lecture on transformer architecture. Start with one task, one tool, one outcome.

How Do You Remove Each Barrier?

Each of the four barriers calls for a different intervention. A blanket "AI training programme" won't address all of them. Here's a practical breakdown.

BarrierShareRoot causeFirst action
Not sure where to start24.2%No clear first taskMap each role's top 3 AI-ready tasks; share worked examples
No time to learn12.1%Workload doesn't allow itProtect 1-2 hours per sprint for AI experimentation
Don't trust the output8.8%Accuracy concerns (often justified)Teach verification workflows; pair AI with human review
My org doesn't support it3.3%Policy, tooling, or culture blocksAudit which tools are approved; publish clear usage guidelines

For the "Not Sure Where to Start" Group

Give them a task, not a course. Pick one repeatable task from their role (drafting a status update, summarising meeting notes, generating test data) and walk them through it with a specific tool. The prompting skills guide provides a structured starting point for writing effective prompts.

For the "No Time to Learn" Group

Make learning indistinguishable from working. Instead of carving out separate training time, embed AI into an existing workflow. Ask them to use AI for one task they're already doing this week. If the result is faster or better, the time argument resolves itself.

For the "Don't Trust the Output" Group

Validate their concern, then equip them. Show them how to verify AI output against known sources, how to spot hallucinated citations, and when to reject a result entirely. These are critical thinking skills that make AI use safer for everyone.

For the "My Org Doesn't Support It" Group

This is a leadership problem. If 3.3% of people feel blocked by their organisation, the fix is structural: approved tool lists, clear data handling policies, and visible executive support. Posts like Is My Company Behind on AI? [2026 Survey] can help you benchmark where your organisation stands.

Measuring Whether Barriers Are Actually Falling

Removing barriers is only useful if you can tell it's working. Adoption metrics like "number of licences provisioned" or "logins per month" measure activity, not capability. A person can log in daily and still produce unreliable outputs.

A more useful signal is measured AI fluency, how well someone can prompt, evaluate, and apply AI to real work. Running a team AI assessment before and after an intervention gives you a concrete baseline and a way to track progress across the dimensions that matter: prompting, critical thinking, technical understanding, workflow integration, and responsible use.

The AI Fluency Levels: 5 Tiers Explained [2026] post breaks down what each tier looks like in practice, so you can set realistic targets for your team.

About this research

This finding comes from AISA's own research. AISA asks a short research question of people as they start the AISA AI fluency assessment, one question at a time. The question in the chart above was asked from 6 August to 10 August 2026, worded exactly as shown and with the answers in the order shown, and 91 people answered it. Each person counts once: when someone answered more than once, their latest answer is the one counted.

More findings from this research are collected in AISA research and data.

How to cite this research: AISA (2026). "Barriers to AI Adoption at Work [2026 Data]". AISA research. https://aisa.to/blog/barriers-to-ai-adoption-at-work


Related reading: Why Don't People Use AI More? [2026 Data], the companion post exploring how often people actually use AI.

Related reading: AI Skills by Motivation: Why You Learn Predicts Your Score, how the reason someone learns AI shapes how well they learn it.

Related reading: AI Readiness Assessment Checklist [2026], a practical checklist for teams preparing to adopt AI.

Frequently Asked Questions

What are the barriers to AI adoption?

When AISA asked 91 people "What's the biggest thing holding you back from using AI more?", the four barriers named were: not knowing where to start (24.2%), no time to learn (12.1%), distrust of AI output (8.8%), and lack of organisational support (3.3%). The remaining 51.6% said nothing was holding them back.

Why do employees resist AI?

The AISA data suggests that most workplace AI adoption challenges stem from uncertainty and capacity constraints rather than active resistance. The largest barrier group (24.2%) simply doesn't know where to start, and 12.1% cite lack of time. Only 8.8% point to distrust of output, and just 3.3% say their organisation blocks them.

How do you get a team to use AI more?

Match the intervention to the barrier. For people who don't know where to start, provide role-specific first tasks rather than generic training. For those without time, embed AI into existing workflows. For those who distrust output, teach verification techniques. Measure progress with a team AI assessment to see whether skills are actually improving, not just whether tools are being opened.

Ozan Dagdeviren

Ozan Dagdeviren

Founder of AISA — the AI skills assessment platform used by professionals worldwide to measure, certify, and develop their AI fluency. More about AISA

AISA

Curious about your AI Fluency?

AISA helps you measure, prove and improve your AI skills — free report in a 20–40 minute chat.

The Science Behind AISA

Metropolitan PoliceHarvard UniversityCrowdboticsE.S.E.

In 2026, Anthropic published the AI Fluency Index — the largest empirical study of AI fluency to date, analysing nearly 10,000 conversations. AISA covers 93% of the behaviours Anthropic identified as markers of AI fluency and goes even deeper with 4 additional criteria. The U.S. Department of Labor's AI Literacy Framework (TEN 07-25) defines what every worker needs to know about AI — AISA covers 100% of its 25 sub-competencies.Read our analysis: Anthropic's AI Fluency Study & AISA · DOL AI Literacy Framework & AISA

AISA's framework is developed by a team with deep roots in tech, behavioural science, and AI product leadership — the rubric is informed by backgrounds spanning the Metropolitan Police, Harvard, Crowdbotics (Silicon Valley), and the European School of Economics.