# AI At Work: Adoption, Friction, and Workforce Redesign **Author:** G2 Research **Published:** August 2026 **Based on:** This research draws on 359 in-depth interviews with business professionals representing a wide mix of roles, industries, and company sizes. **Topics covered:** AI Adoption, Workforce & Org Design AI has moved into day-to-day operations, but mostly in bounded, supervised ways. With **68%** of respondents reporting moderate operational use and **47%** saying AI is concentrated in workflow-specific support, the dominant pattern is not full autonomy but targeted adoption where outcomes are easier to control. That is already changing work: **65%** said AI is increasing both throughput and time for higher-value activity. The main friction is not technical readiness alone; it is trust, governance, and workforce readiness. **66%** cited privacy, governance, and human-validation guardrails as the biggest barrier, while **91%** discussed the need for continued human oversight. At the same time, adoption is culturally uneven: **37%** see a clear split between enthusiasts and resisters, and resistance is driven mainly by familiarity gaps and inertia (**66%**). In response, leaders are leaning on transparent communication (**60%**) and more explicit AI skill signaling in hiring (**46%**). As productivity gains compound, workforce redesign is following, with **50%** already reporting selective restructuring or headcount reduction. **Executive answer:** AI Has Reached Operational Use ## Additional key findings - Use Is Narrow, Not Enterprise-Wide - Guardrails Outweigh Technical Barriers - Productivity Gains Are Changing Workforce Design **On this page:** [Headline statistics](#headline-statistics) · [What We Found](#what-we-found) · [AI Has Moved Into Everyday Operations, But in Narrowly Defined Ways](#ai-has-moved-into-everyday-operations-but-in-narrowly-defined-ways) · [Most Teams Have Reached Moderate, Workflow-Level AI Adoption](#most-teams-have-reached-moderate-workflow-level-ai-adoption) · [AI use clusters in targeted workflows, not broad productivity](#ai-use-clusters-in-targeted-workflows-not-broad-productivity) · [The Strategic Opening Is to Build an AI-Ready Workforce Without Sacrificing Human Judgment](#the-strategic-opening-is-to-build-an-ai-ready-workforce-without-sacrificing-human-judgment) · [AI boosts output while shifting time to higher-value work](#ai-boosts-output-while-shifting-time-to-higher-value-work) · [AI Hiring Expectations Split Sharply Between Signals and Screening](#ai-hiring-expectations-split-sharply-between-signals-and-screening) · [AI Fluency Dominates, but Adaptability and Judgment Differentiate Success](#ai-fluency-dominates-but-adaptability-and-judgment-differentiate-success) · [Middle Management Persists as a Leaner, AI-Augmented Leadership Layer](#middle-management-persists-as-a-leaner-ai-augmented-leadership-layer) · [Real Adoption Friction Comes From Control, Trust, and Readiness Gaps](#real-adoption-friction-comes-from-control-trust-and-readiness-gaps) · [AI speeds drafting, but human review remains nonnegotiable](#ai-speeds-drafting-but-human-review-remains-nonnegotiable) · [Governance barriers dominate, while readiness challenges span all segments](#governance-barriers-dominate-while-readiness-challenges-span-all-segments) · [AI culture splits sharply between enthusiasts and cautious resisters](#ai-culture-splits-sharply-between-enthusiasts-and-cautious-resisters) · [Familiarity gaps and change inertia fuel most AI resistance](#familiarity-gaps-and-change-inertia-fuel-most-ai-resistance) · [Perceptions split: some see little deskilling, others fear erosion](#perceptions-split-some-see-little-deskilling-others-fear-erosion) · [Leaders Are Using Communication and Skill Signaling to Stabilize Adoption](#leaders-are-using-communication-and-skill-signaling-to-stabilize-adoption) · [Transparent reassurance defines how leaders calm AI job fears](#transparent-reassurance-defines-how-leaders-calm-ai-job-fears) · [AI Is Reshaping Roles Faster Than It Resolves Workforce Design Questions](#ai-is-reshaping-roles-faster-than-it-resolves-workforce-design-questions) · [AI prompts selective cuts while expanding specialized organizational roles](#ai-prompts-selective-cuts-while-expanding-specialized-organizational-roles) · [Cross-cutting themes](#cross-cutting-themes) · [What Leaders Should Do Next](#what-leaders-should-do-next) · [Questions This Research Answers](#questions-this-research-answers) · [Conclusion](#conclusion) · [Methodology](#methodology) --- ## Headline statistics - **68%** — AI Has Reached Operational Use - **47%** — Use Is Narrow, Not Enterprise-Wide - **66%** — Guardrails Outweigh Technical Barriers - **50%** — Productivity Gains Are Changing Workforce Design --- ## What We Found AI has moved into day-to-day operations, but mostly in bounded, supervised ways. With **68%** of respondents reporting moderate operational use and **47%** saying AI is concentrated in workflow-specific support, the dominant pattern is not full autonomy but targeted adoption where outcomes are easier to control. That is already changing work: **65%** said AI is increasing both throughput and time for higher-value activity. The main friction is not technical readiness alone; it is trust, governance, and workforce readiness. **66%** cited privacy, governance, and human-validation guardrails as the biggest barrier, while **91%** discussed the need for continued human oversight. At the same time, adoption is culturally uneven: **37%** see a clear split between enthusiasts and resisters, and resistance is driven mainly by familiarity gaps and inertia (**66%**). In response, leaders are leaning on transparent communication (**60%**) and more explicit AI skill signaling in hiring (**46%**). As productivity gains compound, workforce redesign is following, with **50%** already reporting selective restructuring or headcount reduction. **68% — AI Has Reached Operational Use** Most organizations are beyond pilot mode, with AI now embedded in practical day-to-day workflows. The strategic shift is from experimentation to managing adoption at scale. **47% — Use Is Narrow, Not Enterprise-Wide** Nearly half described AI as workflow-specific support rather than broad productivity infrastructure. That shows adoption is advancing fastest where tasks are bounded and outcomes are easier to supervise. **66% — Guardrails Outweigh Technical Barriers** Privacy, governance, and human-validation concerns are the biggest obstacle to broader use. Organizations are being slowed less by access to tools than by the need to use them safely and responsibly. **50% — Productivity Gains Are Changing Workforce Design** Half of respondents already report selective restructuring or headcount reduction linked to AI. The impact is no longer theoretical; it is beginning to reshape roles, teams, and management expectations. --- ## AI Has Moved Into Everyday Operations, But in Narrowly Defined Ways AI is no longer experimental for many teams. Adoption has reached moderate operational use, is concentrated in workflow-specific support, and is already changing how time is allocated toward both output and higher-value work. > We do have like, automated workflows for our IT support our HR support instances. So we have an AI bot that actually manages the intake so that if you do need a live agent, the AI bot actually evaluates your request first. Director of IT Operations Moderate operational use --- ## Most Teams Have Reached Moderate, Workflow-Level AI Adoption **68%** — of respondents were in moderate operational use of AI AI adoption is now operating at a moderate level across most organizations, with 68% describing active, practical use in day-to-day workflows. Another 23% sit in early-to-moderate adoption, while only 9% remain in early-stage experimentation, showing that most teams have moved beyond isolated pilots into more repeatable deployment. Operational maturity is showing up unevenly across functions. More advanced teams describe AI embedded in support workflows and content production, while the next 23% are transitioning from experimentation into targeted use cases with clearer value. The smallest group remains constrained by regulation, privacy, or uncertainty, highlighting that governance still shapes how quickly adoption scales. **Key takeaways:** - **Most teams are beyond experimentation**: 68% report moderate operational AI use, while only 25% remain in very early experimentation or limited operational use - **Workflow-level adoption is now the norm**: 56% say AI is in moderate use across selected workflows, making this the clear center of maturity - **Maturity is widespread but not yet strategic**: 35% are in early-to-mid adoption with broad but immature access, while 20% are moderately advanced but say AI is not yet central to strategy **Strategic implication:** Shift go-to-market and enablement away from AI awareness toward workflow-specific value delivery: package offerings by use case, prove ROI in a small set of high-frequency processes, and provide implementation support that moves teams from fragmented adoption to repeatable operating models. Price and message for a moderate-maturity market—emphasize integration, governance, and measurable productivity gains—while reserving premium strategy-led offerings for the smaller segment ready to make AI core to enterprise direction. > So for copywriters, they are absorbing more volume of the same tasks. I would say we probably are running now at about four to five x the output from our copywriting team, through the use of AI. Head of Marketing, Credit Ratings and Risk Data Moderate operational use > I would say we, I I placed it probably like a like a three. We are kind of moving beyond the current experiment and experimentation phase into, like, more of a targeted production where we can have specific use cases that can deliver value. Senior Technical Product Lead and Lead Automation Integration Engineer, Retail Early-to-moderate adoption --- ## AI use clusters in targeted workflows, not broad productivity **47%** — of respondents described AI use as support or workflow-specific AI adoption is concentrated in targeted operational workflows rather than general productivity. Nearly half of respondents, 47%, described support or workflow-specific use cases, compared with 32% citing broad cross-functional productivity support. Only one in five pointed to primarily administrative or documentation-focused uses, showing organizations are prioritizing embedded task execution over lighter assistance. This pattern suggests AI is gaining traction where outcomes are easier to define and automate, such as service interactions, invoice handling, and quality control. Broad enterprise use is present but secondary, while administrative support appears more limited and often acts as an entry point rather than the main value driver. In practice, AI deployment is skewing toward function-specific tools tied to measurable workflow improvements. **Key takeaways:** - **AI stays tied to targeted workflows**: 71% report several distinct functional use cases, while just 12% have AI embedded across multiple workflows and applications - **Back-office support dominates current adoption**: 47% use AI across multiple back-office productivity workflows versus 21% using it broadly across internal operations and 14% for single-task admin or documentation support - **Broad productivity remains the exception**: 17% limit AI to role-specific use and only 12% have expanded it across multiple workflows, reinforcing that adoption is concentrated in narrower, workflow-specific support **Strategic implication:** Package and sell AI around workflow-specific outcomes, not enterprise-wide productivity promises. Prioritize repeatable back-office and functional use cases with fast deployment, clear ROI, and role-aligned integrations, then expand through adjacent workflow bundles rather than broad platform rollouts. Sharpen messaging around task acceleration, documentation, and operational support; align pricing to per-workflow or per-team adoption; and build expansion plays that convert proven point use cases into multi-function penetration. > Two areas, really. One is accounts payable. So the matching and paying of invoices. The second being customer services. So the initial interaction that a customer would have with us would be through the AI, chatbot. Procurement Director, Healthcare Support or workflow-specific use cases --- ## The Strategic Opening Is to Build an AI-Ready Workforce Without Sacrificing Human Judgment Organizations that intentionally develop AI fluency alongside judgment, redesign hiring around practical AI capability, and prepare managers to lead AI-enabled teams are best positioned to convert productivity gains into durable workforce value. > Overall, the trend leans toward a hybrid outcome. Where employees are both increasing productivity and dedicating more time to innovation and decision making rather than, routine execution. Research Participant Both higher throughput and higher-value work --- ## AI boosts output while shifting time to higher-value work **65%** — of respondents said AI is driving both higher throughput and more time for higher-value work AI is shifting employee time toward a dual benefit: about two-thirds of respondents said it is delivering both higher throughput and more capacity for higher-value work. Another 29% described gains that are primarily about efficiency and output, while only 6% said the main effect is freeing time mainly for higher-value activities. In practice, AI is reducing manual preparation, reporting, and routine execution so employees can focus more on judgment, problem solving, and decision making. The pattern suggests most organizations are not just doing the same work faster; they are also redirecting effort into more strategic tasks, although a meaningful minority still experience AI chiefly as a productivity tool. **Key takeaways:** - **AI shifts time to higher-value work**: 65% say AI is driving both higher throughput and more time for higher-value work, showing impact goes beyond simple efficiency gains - **Most see gains on both fronts**: 57% report AI is delivering both more volume and higher-value work, far outpacing the 12% who see only a little of both and the 6% who say results depend on use case or implementation stage - **Strategic work is the main destination**: 35% say freed time is shifting to strategic or creative work and 34% to higher-value or frontline work, versus just 13% who say time is merely redirected to other tasks generally **Strategic implication:** Package AI around workforce redesign, not just productivity gains: price and message offerings on dual outcomes—higher throughput plus greater strategic capacity. Prioritize deployments that explicitly route saved time into creative, customer-facing, and decision-oriented work, with operating plans, role expectations, and manager KPIs tied to that reallocation. Segment customers by implementation maturity, offering scale-focused plays for early adopters and higher-value workflow transformation for organizations ready to redesign jobs and team structures. > because employees are spending less time on manual data preparation and recurring reports, can spend more time on problem solving, stakeholder engagement, making decisions, So in some cases, they've absorbed additional scope, but Stephanie changed the work more towards judgment and and decision making. AI Enablement and Revenue Operations Lead, Cloud Software Both higher throughput and higher-value work > I would say, for instance, our MBRs took three weeks to produce, and now using AI tools, Agentek AI tools and work flows, that are internal tools. We're able to produce those reports within a week. Manager / Chief of Staff to Director, Cloud Computing Mostly efficiency and throughput gains --- ## AI Hiring Expectations Split Sharply Between Signals and Screening **46%** — had explicit AI screening or role-specific AI requirements in hiring Hiring is increasingly testing for AI capability, but expectations are far from standardized. Nearly half, 46%, reported explicit AI screening or role-specific AI requirements, while 28% said AI skills are preferred or informally assessed. Just over one-quarter, 26%, said they have no formal AI hiring requirement at all. This split suggests a market moving toward AI readiness through different levels of formality. Some employers are embedding AI directly into job descriptions and scoring systems, while others treat it as a baseline preference rather than a must-have. At the same time, a meaningful minority still lacks policies or hiring criteria, creating uneven expectations across roles and organizations. **Key takeaways:** - Nearly half now formalize AI in hiring: 46% report explicit AI screening or role-specific AI requirements, showing AI expectations are moving from ad hoc interest into the hiring process - AI signals are common, but hard requirements are not universal: 39% show no AI hiring signal at all, while 18% rely only on informal preference or interview probing and 32% say AI is preferred but not required - AI expectations split between role design and candidate screening: 38% limit AI expectations to specific roles, just 2% tier them by role family, and 35% have explicit formal screening or job-design changes tied to AI **Strategic implication:** Segment offerings and GTM by hiring maturity: position low-friction enablement and manager toolkits for the 39% with no AI signal and the 18% using only informal probing, while packaging assessment frameworks, role-specific workflows, and compliance-ready screening support for the 46% with formalized expectations. Price in tiers that map to ad hoc, preferred, and required AI use, and anchor messaging on faster hiring, clearer role design, and reduced screening ambiguity rather than generic AI transformation. > These are more like preferred skills because you know, we definitely very definitely, you know, in increasingly looking for AI literacy as a baseline capability, particularly an understanding of how to work effectively with AI tools. Senior Transformation Leader, Global Technology AI skills preferred or informally assessed > We have not. We have nothing in place currently because we have no, AI policy within the organization currently. Director of HR, Localization Services No formal AI hiring requirement --- ## AI Fluency Dominates, but Adaptability and Judgment Differentiate Success **98%** — of respondents discussed skills employees are being told to build for the AI era AI fluency is the leading skill employees are being told to build for the AI era, cited in 56% of responses. Nearly all respondents, 98%, discussed this topic at all, showing how widely workforce expectations are shifting toward practical AI use, alongside a continued emphasis on human judgment and broader career resilience. Human capabilities and adaptability still matter almost as much in combination: 22% highlighted judgment, communication, and business sense, while another 22% pointed to adaptability and continuous learning. In practice, employers are not treating AI skill as tool mastery alone; they want people who can apply AI meaningfully, question outputs, and keep evolving as roles change. **Key takeaways:** - **Adaptability edges out every other AI-era skill**: 59% say employees are being told to build adaptability and continuous learning, slightly ahead of 55% who emphasize applied AI use and experimentation - **AI fluency is expected, but judgment differentiates**: 55% highlight hands-on AI use and experimentation, while 25% specifically call out AI oversight, deployment, and judgment as the higher-order capability - **Human and business judgment remain critical complements**: 25% emphasize business and strategic judgment and 12% point to human-centered soft skills, showing success in the AI era extends beyond tool comfort alone **Strategic implication:** Shift enablement from AI-literacy training alone to role-based capability building that pairs hands-on AI experimentation with decision quality, change readiness, and business judgment. Package offerings in tiers: foundational AI fluency, workflow-level applied use, and advanced oversight/governance for managers and high-impact roles. Position value around faster adaptation and better decisions—not just tool adoption—and equip frontline teams with continuous learning routines, scenario-based practice, and human-centered communication skills. > Judgment. So, specifically, the ability to supervise or question or improve or enhance AI generated work Executive Director of America's Commercial Operations Human judgment, communication, and business sense > Adaptability isn't about mastering tools. It's about cultivating a growth mindset that keeps you relevant no matter how the landscape shifts. Research Participant Adaptability and continuous learning --- ## Middle Management Persists as a Leaner, AI-Augmented Leadership Layer **70%** — of respondents said middle management remains important and will evolve Middle management is expected to endure in AI-enabled organizations, with 70% saying the role remains important and will evolve. Another 17% expect a leaner layer rather than elimination, while only 13% see the function at real risk. The dominant view is adaptation, not disappearance. The role is shifting away from reporting, coordination, and administrative oversight toward coaching, strategy translation, change leadership, and responsible AI adoption. That creates a clear tension: organizations may need fewer managers overall, but those who remain will carry broader spans, higher accountability, and more people leadership responsibility. **Key takeaways:** - **Middle management remains firmly in place**: 70% say the layer remains important and will evolve, showing broad confidence that AI will reshape management rather than eliminate it - **AI is seen as an enhancer, not a replacement**: 42% say middle management will be augmented not replaced, while 29% say it is strongly protected by human oversight needs and another 13% say it is protected if AI is used well - **The role is shifting upward as layers get leaner**: 53% expect middle managers to move toward higher-value work, while 26% see some tasks exposed but the role remaining and 21% expect a leaner layer with greater structural exposure **Strategic implication:** Design AI offerings around middle managers as the primary control layer: automate reporting, coordination, and routine decision support, while elevating manager-facing tools for coaching, exception handling, risk oversight, and cross-functional alignment. Price and package around productivity lift and span-of-control expansion, not headcount removal. Message AI as a leadership amplifier that creates leaner, higher-value management layers, and target change programs toward manager upskilling, workflow redesign, and governance adoption. > I see that their role increasingly becomes about translating strategy into actions, managing change, ensuring responsible AI adoption, supporting teams as world becomes more AI augmented. Senior Transformation Leader, Global Technology Middle management remains important and will evolve > Middle management I wouldn't say would disappear, but I would say it gets thinner and more accountable. Executive Director of America's Commercial Operations Middle management may become leaner but still needed --- ## Real Adoption Friction Comes From Control, Trust, and Readiness Gaps AI still requires heavy human refinement. Governance and privacy guardrails slow broader use, workplace cultures remain polarized between enthusiasts and resisters, and familiarity gaps, inertia, and concerns about weakened foundational skills continue to limit readiness. > Don't let the AI do your thinking, but let it do the 80% of the repetitive work, low value added work that you were pushing back on or you didn't have enough time to do. Senior Director of Technology Tightly reviewed human-in-the-loop assistant --- ## AI speeds drafting, but human review remains nonnegotiable **91%** — of respondents discussed the degree of human oversight required Human oversight remains central to enterprise AI use: 91% of respondents discussed how much review is needed, and the split was nearly even between AI as a drafting assistant with human refinement, 53%, and a tightly reviewed human-in-the-loop model, 47%. The dominant pattern is augmentation, not unattended automation. Roughly half of leaders position AI as a way to remove repetitive work while preserving human judgment, and the other half emphasize formal review checkpoints, especially in higher-risk contexts. This balance reflects a practical tradeoff: AI can accelerate first drafts and routine tasks, but organizations still depend on people to validate outputs, explain reasoning, and make final decisions. **Key takeaways:** - Human oversight is still mandatory: 91% discussed review requirements, with 50% saying close human review is required and another 21% saying human judgment is essential for final decisions - AI is mainly a support, not a substitute: 44% describe it as a primarily human-directed support tool, 35% use it as a drafting or triage partner, and only 20% say it handles some routine first-pass work - Most teams accept AI for drafting, not finalizing: 27% say light human refinement is enough, but 71% still require either close human review or final human judgment before decisions are made **Strategic implication:** Position AI as a supervised drafting layer, not an autonomous decision-maker, and design workflows that require human signoff on all consequential outputs. Package offerings around review intensity—light refinement for low-risk tasks, close expert review for standard use cases, and mandatory final human judgment for high-stakes decisions. Lead messaging with speed-to-draft and triage efficiency, while pricing premium tiers on governance, auditability, and expert validation rather than full automation. > In the regulatory industry, it's always, like, we don't rely on full automation. There's always a human handoff or a human, step in between where you have to cross check things. Director of AI, Asset Management Tightly reviewed human-in-the-loop assistant > The people that were doing that are definitely doing higher value work and in some cases, different work. We still rely on using a human in the loop to ensure that the decisions made by AI or by tooling are appropriate Director of Service Transformation Drafting assistant with human refinement --- ## Governance barriers dominate, while readiness challenges span all segments **66%** — of those citing AI-adoption barriers pointed to privacy, governance, and human-validation guardrails Privacy, governance, and human-validation guardrails are the leading barrier to effective AI adoption, cited by 66% of respondents. That far exceeds training, buy-in, and usage gaps at 24%, while data, integration, and system-access constraints trail at just 10%, showing that oversight concerns outweigh technical and change-management issues for most organizations. Governance challenges center on building ethical frameworks, defining policies, and ensuring employees can detect bias, hallucinations, and sensitive-data risks before acting on outputs. While training and integration still matter, they are secondary to the need for controlled adoption, especially where organizations are still formalizing review processes and deciding how much human validation AI-generated work requires. **Key takeaways:** - Governance barriers clearly lead AI adoption friction: 66% cite privacy, governance, and human-validation guardrails as barriers, and 53% identify infrastructure, governance, and trust constraints as their primary barrier - Readiness challenges cut across every segment: capability, training, and change readiness appear almost evenly as a primary barrier for 36%, a secondary barrier for 31%, and a minor or emerging barrier for 32% - Governance stands apart from other obstacles: infrastructure, governance, and trust constraints are named the primary barrier by 53%, versus just 15% as a secondary barrier and 18% as a minor or emerging barrier **Strategic implication:** Lead with governance-first offers: package privacy controls, approval workflows, auditability, and human-in-the-loop validation as the entry point to AI adoption, and price or scope implementations around risk reduction and compliance outcomes before broader transformation. Segment enablement separately: embed role-based training, change management, and capability building across all customers as a continuous adoption layer, not a one-time rollout, since readiness gaps persist regardless of barrier severity. > We've not yet. We are looking at what an ethical AI use framework would look like for our organization. And looking at the right governance to wrap around that. Director of Finance, Healthcare Board Privacy, governance, and human-validation guardrails > So they need to be able to understand the data sensitivity the bias, the hallucinations, and can validate outputs instead of trusting them blindly. Executive Director of America's Commercial Operations Privacy, governance, and human-validation guardrails --- ## AI culture splits sharply between enthusiasts and cautious resisters **37%** — saw a clear divide between AI enthusiasts and resisters AI culture is notably polarized in many organizations, with 37% reporting a clear divide between enthusiasts and resisters. An equal 37% described broad acceptance with little cultural split, while 26% saw only mild unevenness, suggesting workplaces are just as likely to be unified as sharply divided. Enthusiasts are often early adopters who integrate AI quickly, while resisters are more likely to question data risks, trust, or job impact. That mix creates uneven adoption in practice: some teams move fast with AI-enabled workflows, while others remain cautious, quiet, or hesitant even when organization-level support is strong. **Key takeaways:** - **AI culture is meaningfully polarized**: 61% report at least a moderate split between enthusiasts and resisters, including 20% who see strong polarization and 41% who see a moderate or uneven divide - **Resistance is mostly a readiness gap**: 44% say hesitation is driven primarily by familiarity and readiness, and that it softens with exposure, versus just 8% who describe resistance as entrenched and mindset-based - **The divide is not universal**: 39% report broad acceptance or little to no split, showing polarization is significant but not dominant across all organizations **Strategic implication:** Segment AI rollouts by audience: equip enthusiasts as visible champions and design low-friction onboarding, role-based training, and guided use cases for hesitant groups. Lead messaging with practical workflow gains, risk controls, and peer proof rather than visionary language. Price and package for staged adoption with pilot programs, enablement support, and expansion paths, since most resistance reflects readiness gaps that can be converted with exposure, while preserving targeted change-management for pockets of entrenched skepticism. > So, like, AI natives quickly adapt tools and kind of integrate it into their workflow, whereas AI resistors tend to be skeptical or cautious. Senior Technical Product Lead and Lead Automation Integration Engineer, Retail Clear divide between AI enthusiasts and resisters > Some of our legacy employees don't agree with it or think it's gonna take jobs away and are very hesitant on it. Group Talent Development Manager, Mobility Services Clear divide between AI enthusiasts and resisters --- ## Familiarity gaps and change inertia fuel most AI resistance **66%** — of respondents pointed to mindset inertia and AI familiarity gaps as the main driver of AI resistance AI resistance is driven chiefly by familiarity gaps and entrenched habits, with two-thirds of respondents citing mindset inertia and limited AI understanding as the primary barrier. By comparison, only 8% pointed to fear of job loss, while about one-quarter reported little to no meaningful resistance at all. Resistance appears less rooted in outright opposition than in low trust, uncertain relevance, and comfort with established ways of working. Tenure and organizational history seem to reinforce that inertia, while job-replacement concerns remain a secondary but visible barrier. In practice, adoption efforts should prioritize trust-building, practical exposure, and role-specific use cases over broad reassurance alone. **Key takeaways:** - Combined capability and mindset barriers dominate resistance: 70% cite both limited AI familiarity and change inertia together, far outweighing training gaps alone at 14% or mindset inertia alone at 15% - Fear is secondary to readiness barriers: while 32% point to general fear or uncertainty, only 12% cite job replacement fears and another 12% mention mixed fear concerns, well below the 66% centered on familiarity and inertia - AI resistance is mostly about adoption, not alarm: 66% attribute resistance primarily to familiarity gaps and mindset inertia, showing the biggest barrier is learning and behavior change rather than direct job security concerns at 12% **Strategic implication:** Prioritize change enablement over fear-based reassurance: package AI rollouts with role-specific onboarding, hands-on use cases, manager-led reinforcement, and simple workflow integration to close both familiarity and mindset gaps. Position pricing and packaging around guided adoption—training, implementation support, and success milestones—rather than standalone licenses. Tailor messaging to productivity, confidence, and ease of use, while addressing job-loss concerns as a secondary barrier rather than the primary sales narrative. > the trust perceived relevance and I think how long you've been with the organization. Like, things were working a certain way. Why do you really need to change it? Director of Data Analytics Mindset inertia and AI familiarity gaps > There's definitely a mindset issue where there are people who feel their jobs are threatened by AI, and you can see their resistance and their reluctance to both adopt it and also accept it. Group Product Manager, Technology Fear and job-replacement anxiety --- ## Perceptions split: some see little deskilling, others fear erosion **39%** — of respondents expressed concern about erosion of foundational skills Concern about AI weakening foundational skill development is the largest single viewpoint, cited by 39% of respondents. Still, the picture is not settled: 37% report no current deskilling concern or say it is too early to tell, while 24% believe AI can strengthen learning when paired with human guidance and oversight. The divide reflects a practical tension between speed and skill formation. Respondents raising concerns describe juniors relying on AI for answers without mastering basics or explaining their work, while others say structured training and critical review can prevent overreliance. In practice, organizations appear split between early warning signs and confidence that safeguards can preserve core capabilities. **Key takeaways:** - Perceptions are genuinely split: 41% report no deskilling observed, while 39% express concern about erosion of foundational skills, showing no clear consensus on AI’s impact - Most skepticism is cautious, not dismissive: beyond the 41% seeing no deskilling, 17% say it is too early to tell and 15% describe concerns as limited or isolated rather than systemic - Overreliance fears center on capability building: 30% worry AI is reducing verification and creating overreliance, while 27% specifically fear junior employees are missing foundational skill development **Strategic implication:** Segment offerings by risk posture: position AI as a productivity accelerator for teams seeing little deskilling, while packaging governance, verification workflows, and skill-safeguard features for buyers worried about erosion. Build messaging around “assist, then validate,” and price premium tiers around auditability, review controls, and training programs for junior staff. In deployment, require human-check checkpoints, benchmark foundational competencies, and track early-career skill development to prevent overreliance from becoming organizational weakness. > You know, as an organization, so we've been addressing this by reinforcing you know, foundation skills, foundational skills, you know, encouraging critical review of AI outputs, and positioning AI as a support tool rather than a replacement for learning or judge judgment, if that makes sense. Senior Transformation Leader, Global Technology AI can support learning if balanced with human input --- ## Leaders Are Using Communication and Skill Signaling to Stabilize Adoption In response to anxiety and uneven readiness, leaders are managing the transition through transparent communication, increasingly embedding AI expectations into hiring, and signaling that AI fluency must be paired with judgment and adaptability. > Teams are informed on which tax are gonna be automated, how the workflows will change, the new skills that will become valuable, and we would provide know, training opportunities for them as well. Lead of the Data Analytics and Technology Organization Globally Transparent communication about AI and job impact --- ## Transparent reassurance defines how leaders calm AI job fears **60%** — communicated transparently about AI and its impact on jobs Leaders most often manage AI-related workforce anxiety through openness. Three in five said they communicate transparently about AI and its job impact, while 22% pair that message with explicit reassurance that AI is intended to augment work. By contrast, fewer than one in five keep communication quiet or limited. The strongest approaches combine clarity on what will change with reassurance about why it is changing. Transparent leaders explain which tasks will be automated, how roles may shift, and what support employees will receive; others soften concern by positioning AI as a productivity tool rather than a headcount strategy. The smaller quiet segment risks leaving uncertainty to fill the gap. **Key takeaways:** - Transparent communication is the dominant anxiety strategy: 60% say leaders communicated transparently about AI and its impact on jobs, including 57% who describe communication as proactive and transparent versus 17% who say it was quiet, selective, or deferred - Reassurance outweighs explicit workforce warnings: 44% say leaders stressed no replacement or continuity, compared with 18% who openly acknowledged workforce change or headcount pressure - Most leaders pair openness with optimism, but depth still varies: while 57% report proactive transparency, another 24% say communication is transparent but still limited or early-stage, and 25% say leaders framed AI around upskilling, redeployment, and career growth **Strategic implication:** Lead with a structured AI workforce narrative: publish role-by-role impact updates, pair every automation announcement with clear continuity, upskilling, and redeployment pathways, and equip managers with repeatable talking points for team discussions. Segment support by communication maturity—foundational transparency packages for early-stage organizations and deeper workforce-transition planning for advanced adopters. Position offerings around change readiness and trust-building, and price premium services around manager enablement, reskilling design, and workforce scenario planning. > We regularly share updates about our automation strategy and emphasize that AI is meant to augment, not to replace human talent. Research Participant Transparent communication about AI and job impact > we frame AI as a productivity and augmentation strategy first, not a headcount reduction strategy, definitely. Senior Transformation Leader, Global Technology Reassurance framing AI as augmentation --- ## AI Is Reshaping Roles Faster Than It Resolves Workforce Design Questions These adoption patterns and coping mechanisms are already affecting organizational design: some firms report selective restructuring or headcount reduction, while middle management is expected to persist in a leaner, more AI-enabled leadership role. > Instead of expanding teams linearly with workload, we are now able to maintain or even reduce headcount in certain operational areas while reallocating resources toward higher value strategic roles. Research Participant Selective restructuring or headcount reduction --- ## AI prompts selective cuts while expanding specialized organizational roles **50%** — reported selective restructuring or headcount reduction AI is reshaping workforce models through targeted rather than broad-based cuts: 50% reported selective restructuring or headcount reduction, while 29% said AI is creating new roles or redefining existing ones. Only one in five said AI has not yet affected headcount or organizational design, underscoring that workforce impact is already material for most organizations. In practice, companies are using AI to reduce labor intensity in operational work while shifting investment toward specialized AI leadership and technical talent. This creates a two-track pattern: efficiency-driven reductions in some functions, alongside net-new roles such as AI officers, digital transformation teams, and AI engineers, while a smaller group is still treating AI primarily as a productivity gain without org redesign. **Key takeaways:** - AI is reshaping teams selectively, not universally: 33% report selective or direct headcount reduction, while 30% say AI has had no current headcount impact - Efficiency gains are often coming without hiring growth: 36% say AI is improving efficiency without increasing headcount, outpacing the 33% seeing direct reductions - New AI roles are expanding faster than structures are standing still: 47% have created net-new AI or coordination roles, compared with 31% absorbing AI into existing roles and 22% making no role or structural changes **Strategic implication:** Segment accounts and operating plans by AI workforce posture: position automation and governance offers for the 33% pursuing selective cuts, productivity and redeployment programs for the 36% driving efficiency without headcount growth, and advisory for the 30% still in wait-and-see mode. Price around measurable labor leverage and speed-to-value, while pairing every deployment with role redesign, reskilling, and net-new AI coordination support, since specialized AI roles are expanding faster than organizations remain structurally unchanged. > Yes. We have introduced a whole new department called digital transformation. Just to manage the AI solution. And also, we started to hire more data scientists and AI engineers. We never had this before. Digital Transformation Lead, Chemicals Net-new AI roles or role redefinition > At this moment in time, it's included as an efficiency gain, so we can do more with same amount of people as opposed to reducing the size of the organization. Chief Operating Officer, Insurance No current headcount or org-design impact --- ## Cross-cutting themes ### Operational Adoption Without Full Autonomy AI has reached moderate operational use and is producing measurable workflow benefits, but its use remains largely workflow-specific because human refinement is still essential and privacy/governance guardrails constrain broader application. Organizations should stop framing success as full automation and instead design for supervised, bounded AI use cases where governance and human review are built into the operating model. ### The Readiness Divide A visible split between AI enthusiasts and resisters is reinforced by familiarity gaps and inertia, which in turn forces leaders to rely on transparent communication and explicit skill-building signals to reduce anxiety and move adoption forward. Closing the adoption gap requires treating AI change as a workforce-readiness challenge, not just a tooling rollout, with communication, training, and clearer expectations working together. ### Productivity Gains Are Converting Into Workforce Redesign As AI shifts employee time toward greater throughput and higher-value work, organizations are beginning to selectively restructure teams and redefine leadership layers, while also increasing emphasis on AI-related hiring and future-oriented skills. The strategic question is no longer whether AI changes work, but how deliberately organizations redesign roles, management responsibilities, and talent pipelines around that change. --- ## What Leaders Should Do Next ### Design for supervised AI, not full automation Prioritize workflow-specific use cases where human review is built into the process, since AI adoption is already strongest in bounded tasks and 91% of respondents emphasized ongoing human oversight. Treat governance and refinement as part of the operating model rather than as temporary friction. ### Treat adoption as a workforce-readiness challenge Close resistance by investing in practical AI familiarity, role-based training, and clearer expectations, because 66% linked resistance to familiarity gaps and inertia rather than job-loss fear. Support this with transparent communication about what will change and what will remain human-led. ### Redesign jobs around judgment plus AI fluency Update job descriptions, hiring criteria, and performance expectations to reflect the emerging combination of AI fluency, judgment, and adaptability. This is increasingly urgent as 46% already use explicit AI screening and 56% are telling employees to build AI fluency. ### Protect foundational skill development while scaling AI use Build coaching, review checkpoints, and learning standards into AI-enabled work so speed does not come at the expense of core capability. This is important because 39% expressed concern that AI could erode foundational skills, especially for less experienced employees. ### Prepare managers for leaner, AI-enabled leadership roles Invest in middle-manager capability for coaching, change leadership, and responsible AI adoption, since workforce redesign is already underway and 70% expect middle management to remain important but evolve. The manager role should shift from administrative oversight toward judgment, coordination, and team enablement. --- ## Questions This Research Answers **How far has AI adoption really progressed inside organizations?** AI adoption has moved well beyond isolated pilots for most organizations. A full 68% reported moderate operational use in day-to-day workflows, another 23% described early-to-moderate adoption, and only 9% remain in early-stage experimentation. **Is AI being used broadly across work, or mainly in specific tasks?** Use is concentrated in narrower operational applications. While 32% described broad cross-functional productivity support, the largest share, 47%, said AI is mainly used for support or workflow-specific tasks, and only 20% pointed primarily to administrative or documentation use. **What is the biggest thing slowing more effective AI adoption?** The biggest barrier is not basic access to tools but control and oversight. Among those citing barriers, 66% pointed to privacy, governance, and human-validation guardrails, compared with 24% citing training, buy-in, and usage gaps and just 10% citing data, integration, or system-access issues. **Why are some employees resisting AI?** Resistance is driven mainly by familiarity gaps and inertia, not widespread fear of job loss. Two-thirds, 66%, said mindset inertia and limited understanding are the primary drivers, while only 8% pointed to job-loss fears; about one-quarter reported little to no meaningful resistance. **Is AI already affecting workforce structure and management?** Yes. Half of respondents, 50%, said AI is driving selective restructuring or headcount reduction, while 29% reported the creation of new roles or redefinition of existing ones. At the same time, 70% believe middle management remains important and will evolve rather than disappear. --- ## Conclusion ### Design for supervised AI, not full automation Prioritize workflow-specific use cases where human review is built into the process, since AI adoption is already strongest in bounded tasks and 91% of respondents emphasized ongoing human oversight. Treat governance and refinement as part of the operating model rather than as temporary friction. ### Treat adoption as a workforce-readiness challenge Close resistance by investing in practical AI familiarity, role-based training, and clearer expectations, because 66% linked resistance to familiarity gaps and inertia rather than job-loss fear. Support this with transparent communication about what will change and what will remain human-led. ### Redesign jobs around judgment plus AI fluency Update job descriptions, hiring criteria, and performance expectations to reflect the emerging combination of AI fluency, judgment, and adaptability. This is increasingly urgent as 46% already use explicit AI screening and 56% are telling employees to build AI fluency. ### Protect foundational skill development while scaling AI use Build coaching, review checkpoints, and learning standards into AI-enabled work so speed does not come at the expense of core capability. This is important because 39% expressed concern that AI could erode foundational skills, especially for less experienced employees. ### Prepare managers for leaner, AI-enabled leadership roles Invest in middle-manager capability for coaching, change leadership, and responsible AI adoption, since workforce redesign is already underway and 70% expect middle management to remain important but evolve. The manager role should shift from administrative oversight toward judgment, coordination, and team enablement. The research shows a clear transition from AI experimentation to operational adoption, but not to fully autonomous work. Instead, organizations are settling into a model of operational adoption without full autonomy: 68% report moderate day-to-day use, 47% say that use is concentrated in workflow-specific support, and 65% report that AI is improving both throughput and capacity for higher-value work. The center of gravity has shifted from asking whether AI is useful to deciding where it can be used safely, effectively, and with the right level of human judgment. That progress is constrained by a readiness divide. Privacy, governance, and human-validation guardrails are the top barrier for 66% of respondents, while 91% emphasized that human refinement or formal review remains essential. Cultural adoption is also uneven: 37% see a clear split between enthusiasts and resisters, and resistance is most often tied to familiarity gaps and inertia rather than outright opposition. These tensions help explain why leaders are relying on transparency, reassurance, and skill signaling to stabilize adoption while concerns about foundational skill erosion remain unresolved. The strategic opportunity is to build an AI-ready workforce without sacrificing human judgment. Organizations should standardize bounded, supervised use cases; invest in workforce readiness through training and communication; and redesign hiring, roles, and management expectations around practical AI fluency, adaptability, and decision-making. That matters because productivity gains are already converting into workforce redesign: 50% report selective restructuring or headcount reduction, 46% have explicit AI screening or role-based hiring requirements, and 70% expect middle management to continue in a leaner, more AI-enabled form. The winners will be the organizations that treat AI not as a shortcut to replacing people, but as a catalyst for redesigning work more deliberately. The bottom line: the future belongs not to organizations that automate the most, but to those that combine AI fluency, governance discipline, and human judgment into a better way of working. The bottom line: the future belongs not to organizations that automate the most, but to those that combine AI fluency, governance discipline, and human judgment into a better way of working. --- ## Methodology G2 is the world's largest and most trusted software marketplace. This research draws on 359 in-depth interviews with business professionals representing a wide mix of roles, industries, and company sizes. Interviews ran 1 to 34 minutes and covered Primary AI Use Case Scope, How AI Is Changing Employee Time Allocation, AI Adoption and Maturity Stage, and Primary Barriers to Effective AI Adoption. The conversational format allowed respondents to discuss their actual practices rather than select from preset options, surfacing nuance that closed-ended surveys typically miss. Respondents included business professionals across technology, financial services, healthcare, manufacturing, and retail. All participants were selected for their direct experience with AI adoption and workplace use cases. Company sizes ranged from small businesses to large enterprises. The analysis of 359 interview transcripts was conducted using AI for semantic understanding, with multi-iteration validation and cross-verification to ensure analysis quality. Each transcript was independently reviewed by G2's AI Custom Research team to inform narrative, context, and clarity. G2 Research, June 2026 --- *Source: G2 Research — "AI At Work: Adoption, Friction, and Workforce Redesign" (August 2026). Based on G2 first-party research. Markdown edition structured for answer-engine optimization (AEO).*