Free tools Windows power users keep installed
One-click scans. No signup required.
Workers do not approach workplace AI in one uniform way. Slack’s Workforce Lab grouped respondents into five categories—from vocal, frequent users to skeptics and cautious nonusers—to help explain why a single adoption strategy can miss what employees actually need. The study surveyed 5,000 full-time desk workers, but it is a 2024 snapshot, not a measure of today’s workforce.
Slack identified five ways workers approach AI
Slack’s Workforce Lab said its study of 5,000 full-time desk workers combined quantitative survey research with in-depth interviews about what motivates people to use AI and how they feel about using it at work. The categories describe current behavior and attitudes, not permanent personality types or psychological diagnoses. Slack described them as a snapshot of how workers feel now.
| Persona | Typical behavior or attitude | Likely organizational need |
|---|---|---|
| Maximalist | Uses AI frequently, sees benefits, and openly encourages colleagues to use it. | A safe way to demonstrate useful workflows and teach peers. |
| Underground | Uses AI substantially but keeps that use private or does not promote it. | Clear rules, approved tools, and confidence that permitted use will not be stigmatized. |
| Rebel | Distrusts or rejects AI hype, avoids using it at work, or sees colleagues’ use as unfair. | Credible answers about fairness, job impact, quality, and accountability. |
| Superfan | Is enthusiastic about AI’s potential but has not yet incorporated it meaningfully into work. | Practical, guided first experiences with low-risk tasks. |
| Observer | Watches AI’s development with interest or caution but has not begun using it at work. | Clear information and optional exposure without pressure to adopt. |
The categories overlap in an important way: a Superfan and a Rebel may both be nonusers, but for opposite reasons. One may need practice; the other may need a convincing explanation of how risks and fairness will be handled. Treating both as simply “resistant” would obscure what could actually help.
The reported percentages are not a complete global breakdown
VentureBeat’s September 4, 2024 account of the 5,000-worker study reported 30% Maximalists, 20% Undergrounds, and 19% Rebels. It did not provide exact percentages for Superfans and Observers in the reported figures, so those shares should not be inferred by splitting the remainder. VentureBeat’s account gives the three reported global-study shares.
#1 Best Overall
A separate Slack study in Singapore illustrates why the samples must not be combined. Salesforce said its Singapore survey included 1,031 workers surveyed August 6–14, 2024; the reported shares were 25% Maximalists, 22% Undergrounds, 13% Rebels, 23% Superfans, and 17% Observers. These figures apply to that Singapore sample, not to the 5,000-person study. Salesforce’s announcement describes the Singapore sample and its results.
Why the personas are more useful than an AI user/nonuser split
Frequency of use alone cannot explain whether people feel safe using a tool, understand its limits, or believe its use is fair. Slack’s framework draws attention to several dimensions leaders need to consider:
- Behavior: whether someone uses AI at work and how much.
- Visibility: whether the person tells colleagues or keeps use private.
- Emotion and trust: enthusiasm, anxiety, skepticism, or caution.
- Perceived norms: whether use seems authorized, safe, and fair.
- Readiness: whether the next useful step is permission, explanation, practice, or evidence.
These distinctions are especially valuable when a company is setting expectations. A quiet user may be concerned about confidentiality or judgment, rather than deliberately evading rules. A cautious nonuser may be waiting for guidance, not opposing technology. Conversely, frequent use does not by itself show that the output is accurate, safe, or valuable.
Unclear rules are associated with less experimentation
Slack reported that 37% of desk workers said their company had no AI policy, and that workers at companies without guidelines were six times less likely to have experimented with AI tools than workers at companies with established guidelines. This is an association reported by Slack, not proof that a policy alone causes experimentation. Organizations with guidelines may also provide approved tools, training, leadership support, or more mature technology programs; “no policy” may also mean no formal written policy despite informal norms. A restrictive policy could discourage experimentation rather than enable it. Slack’s Workforce Lab account reports the policy and experimentation findings.
The practical implication is not simply to publish a policy. Workers need to know what is allowed, what information is protected, when a human must check the result, and where to take questions. Those details can reduce uncertainty without treating all hesitation as a problem to overcome.
How employers can respond to each persona
Maximalists: turn individual enthusiasm into shared learning
Give frequent users approved spaces to show how they use AI, share examples, and explain what did not work. Peer demonstrations can help colleagues learn without making adoption depend on informal evangelism. Slack’s suggested response includes creating room for these employees to share their practices. Avoid presenting a Maximalist’s workflow as automatically reliable: generated content still needs appropriate review, and enthusiasm should not outrun privacy, security, or quality controls.
Rank #3
Undergrounds: make legitimate use visible and safe
State which tools are approved, what data may be entered, whether AI assistance must be disclosed, and how employees can ask questions without fear of punishment for seeking clarification. Slack recommended clear permissions and guidelines to help bring private use into the open. Quiet use is a signal to investigate the environment, not evidence by itself of misconduct: employees may be responding to unclear rules, confidentiality concerns, or stigma.
Rebels: address the underlying concerns rather than demanding compliance
Invite skeptical employees to identify failure modes and discuss fairness, attribution, bias, privacy, job impact, and accountability. Show how human review works and where AI is intended to assist rather than replace judgment. Slack suggested training and explaining that AI can support productivity rather than constitute “cheating.” A rollout that treats adoption as inevitable or dismisses dissent risks creating performative compliance and less candid feedback.
The Tool Desk
Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Superfans: turn interest into a useful first experience
Offer guided workshops with realistic tasks, approved tools, and simple ways to check results. Pairing an interested novice with an experienced colleague can make the first attempt more concrete. Enthusiasm is not the same as competence, so teach workers how to verify output and when not to use AI. Assess whether a workflow improves quality or saves time rather than counting logins.
Observers: reduce uncertainty and keep the first step optional
Explain what a tool can and cannot do, demonstrate it on low-consequence tasks, and offer optional training or office hours. Some Observers may be waiting for evidence, policy, or examples from colleagues. Pressure can turn caution into resistance; a clear route to ask questions lets people make an informed choice.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.Use Slack’s PET plan to turn guidance into practice
Slack frames AI enablement around Permission, Education, and Training. The framework is platform-neutral: it can guide an organization using Slack, another collaboration suite, or approved standalone tools. Slack outlines the PET approach.
Permission: make the boundaries usable
- Which tools are approved, and are personal accounts permitted?
- What company, employee, or customer data must not be entered?
- When must employees disclose AI assistance or obtain approval?
- Who is accountable for the final work, and which outputs require human review?
Education: explain capabilities and limits
- What tasks can the tool handle reliably, and where does it commonly fail?
- How should workers check for errors, bias, or unsupported claims?
- What are the organization’s expectations for privacy, security, attribution, and copyright?
Training: practice on real work
- Which workflows are appropriate to augment, and what does a good result look like?
- How should employees verify output, escalate uncertain cases, or revert to a human-led process?
- How will the organization assess accuracy, rework, completion time, user confidence, and safety—not just speed?
A workable rollout also needs feedback and the ability to change course. Leaders should check whether all relevant workers have access to tools and instruction, provide a way to report harms or errors, and be willing to stop a use case that performs badly. Managers matter too: contradictory signals from supervisors can undermine even a clear written policy.
Best Value
What the survey can—and cannot—establish
The findings are attributed to Slack’s Workforce Lab, whose research is sponsored by a company that sells workplace collaboration and AI software. The survey offers a useful way to think about different worker attitudes, but the figures are self-reported and the source coverage does not establish independent replication. The categories are not universal psychological types, and they should not be used to diagnose employees or make employment decisions.
Nor does the study show that AI use necessarily improves work. Organizations should evaluate accuracy, rework, customer or employee impact, human-review rates, security incidents, worker confidence, and whether time saved is put toward higher-value work. Productivity claims need to be treated as outcomes to measure, not assumptions built into a rollout.
The 2024 snapshot is not the current adoption rate
In a later Workforce Index report published June 26, 2025, Slack said 60% of desk workers were using AI and 42% were using it regularly, at least weekly. Slack also reported that daily users described higher productivity, focus, and job satisfaction than workers who had not yet embraced AI. Those are Slack’s survey findings and self-reports, not evidence that AI caused the differences. Slack’s June 2025 report describes those later findings.
The later figures show why the 2024 persona shares should not be presented as current workforce statistics. They do not show that the personas caused adoption to rise. Even as use becomes more common, visibility, confidence, trust, and perceived fairness can still differ from worker to worker.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Repair Windows errors before they cause bigger problems3Fix the driver behind crashes, sound loss and screen glitchesQuick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
