# Guardrails, Not Rejection: How People Really See AI **Author:** G2 Research **Published:** June 2026 **Based on:** This research draws on 121 in-depth interviews with business professionals representing a wide mix of roles, industries, and company sizes. **Topics covered:** AI Adoption, AI Trust & Governance How 121 business professionals are navigating AI adoption and governance: 56% favor progress only with guardrails in place, 59% see AI as complementary to work with selective task automation, 91% say humans must retain final decision authority and accountability, and 98% identified specific groups at risk of being left behind. **Executive answer:** 56% guardrailed progress is the dominant position: Most respondents support moving forward with AI, but only if adoption visibly includes safeguards and acknowledges uneven readiness across society. **On this page:** [Overall Worldview About AI’s Long-Term Trajectory](#overall-worldview-about-ai-s-long-term-trajectory) · [Preferred Pace of AI Adoption and Perceived Social Readiness](#preferred-pace-of-ai-adoption-and-perceived-social-readiness) · [Expected Role of AI in Human Work](#expected-role-of-ai-in-human-work) · [Types of Work Respondents Expect AI to Automate First](#types-of-work-respondents-expect-ai-to-automate-first) · [Job Displacement Expectations and Labor Transition Outlook](#job-displacement-expectations-and-labor-transition-outlook) · [Perceived Winners in the AI Economy](#perceived-winners-in-the-ai-economy) · [Groups Seen as Most at Risk of Being Left Behind](#groups-seen-as-most-at-risk-of-being-left-behind) · [Trust, Truthfulness, and Misuse Risks](#trust-truthfulness-and-misuse-risks) · [Concern About Dependence on AI and Erosion of Human Thinking](#concern-about-dependence-on-ai-and-erosion-of-human-thinking) · [Human Boundaries and Irreplaceable Domains](#human-boundaries-and-irreplaceable-domains) · [Expectations for Human Oversight and Final Accountability](#expectations-for-human-oversight-and-final-accountability) · [Sources of Meaning and Human Purpose in an AI-Heavy Future](#sources-of-meaning-and-human-purpose-in-an-ai-heavy-future) · [Strategic takeaways](#strategic-takeaways) · [Frequently asked questions](#frequently-asked-questions) · [How this was produced](#how-this-was-produced) --- ## Overall Worldview About AI’s Long-Term Trajectory AI is widely seen as an unavoidable force for change, with 55% describing it as inevitable and transformative. Another 43% take a more bounded view, expecting major impact without subscribing to apocalyptic outcomes. Only 2% express outright skepticism or uncertainty about the broader AI narrative. This split points to strong momentum behind adoption, but not blind enthusiasm. While a majority expect AI to reshape daily life and work at scale, a similarly large segment emphasizes human stewardship, tradeoffs, and the limits of extreme utopian or dystopian narratives. In practice, leaders should plan for deep integration while maintaining clear accountability and governance. **Key takeaways:** - AI’s transformation feels inevitable, not speculative: 100% expect AI to be transformative, with 66% seeing change as gradual and 34% expecting rapid disruption - Apocalypse narratives fail to convince most: 97% stay away from extreme hype or doom, including 49% who are strongly pragmatic and 48% who hold mixed but still non-extreme views - Pragmatism edges out polarization: only 3% lean toward extreme narratives, while nearly half are firmly bounded in their outlook at 49% **Strategic implication:** Position AI offerings as essential, staged business transformation rather than revolutionary salvation or existential threat. Lead messaging with practical outcomes, deployment roadmaps, and governance assurances, since most buyers expect meaningful change but reject extremes. Price around scalable adoption tiers and expansion paths that support gradual enterprise rollout, while preserving premium options for faster movers. Equip sales and product teams to anchor conversations in ROI, risk management, and operational integration, not hype-driven disruption narratives. --- ## Preferred Pace of AI Adoption and Perceived Social Readiness Most respondents favor continued AI progress, but with guardrails: 56% support moving forward while acknowledging that social readiness is uneven. Another 36% want a slower, more cautious pace because society is not ready, while only 8% believe adoption should move faster because people and institutions can adapt. This pattern points to a pragmatic middle ground rather than broad enthusiasm or outright resistance. The dominant concern is not AI’s potential, but whether policy, education, workforce development, and public understanding can keep pace. In practice, leaders should keep advancing AI while investing more aggressively in governance, training, and institutional readiness. **Key takeaways:** - Most want AI progress with guardrails: 67% support continuing AI advancement with conditions, and 56% explicitly favor moving forward while recognizing that social readiness is uneven - Caution outweighs calls to accelerate: 38% prefer a gradual rollout and 25% want AI adoption slowed, versus just 17% who prefer faster adoption - Few believe society is fully keeping up: only 9% say society can broadly adapt to the current pace, while 35% are concerned about readiness even if they do not explicitly call for slower adoption **Strategic implication:** Lead with a paced-deployment strategy: launch AI in phased releases, make guardrails visible, and price around trust-building support such as onboarding, governance, and human oversight. Position faster, higher-autonomy offerings as opt-in tiers for more AI-ready segments, while default messaging emphasizes safety, control, and practical adaptation. Invest in readiness enablement—training, change management, and clear use policies—to reduce friction among the larger cautious majority and expand adoption without triggering backlash. --- ## Expected Role of AI in Human Work AI is expected to augment work more often than replace it outright, with 59% of respondents describing a hybrid future where it supports people while taking over selected tasks. Another 36% saw AI mainly as a complement to human work, while only 6% anticipated broad replacement and major job transformation. This pattern points to a pragmatic division of labor: AI is welcomed for repetitive, operational, and productivity-oriented work, while humans retain responsibility for judgment, strategy, and high-stakes decisions. The strongest consensus sits between pure augmentation and full automation, suggesting organizations should redesign tasks and workflows, not assume wholesale job elimination. **Key takeaways:** - AI is seen mainly as an augmenting force: 76% say AI works best with human oversight and judgment, while 59% explicitly view it as both a complement to human work and a selective replacement for specific tasks - Job transformation outweighs job replacement: 94% expect some jobs to be replaced but most work to be transformed, compared with just 3% who foresee broad replacement across many roles - Pure automation is a fringe expectation: only 2% believe AI will only automate tasks, while 23% take a strongly augmentation-first view that centers AI on supporting rather than replacing human work **Strategic implication:** Position AI offerings as human-in-the-loop productivity multipliers, and package automation as targeted modules for repetitive, low-value tasks rather than end-to-end worker replacement. Prioritize workflow redesign, training, and governance features that preserve human judgment, and sell against transformation outcomes such as speed, quality, and capacity expansion. Use value-based pricing tied to efficiency gains, while messaging emphasizes augmentation, role evolution, and selective task automation to reduce adoption resistance. --- ## Types of Work Respondents Expect AI to Automate First Knowledge and analytical work is the leading area respondents expect AI to automate first, cited by 43%, ahead of routine administrative and repetitive tasks at 31%. Another 26% anticipate broader cross-sector disruption, extending beyond office work into customer service, factory roles, and some physical jobs. The pattern suggests respondents are not limiting early automation to low-skill work. Many expect AI to move quickly into structured intellectual tasks such as analysis, reporting, and coding, while still absorbing repetitive admin work in parallel. In practice, this points to earlier pressure on knowledge roles than many organizations may assume. **Key takeaways:** - Analytical work is seen as first in line: 79% expect decision, analytical, or strategic work to be automated early, far ahead of research and information tasks at 2% - Routine administration follows close behind: 54% point to routine admin and clerical tasks as early automation targets, compared with 35% for structured service or workflow tasks - Professional impact is expected to be broad, not just entry level: 19% foresee cross-sector professional roles being affected first, while only 3% expect lower-level or low-interaction roles to lead the shift **Strategic implication:** Prioritize AI products and change programs around high-value analytical workflows first—decision support, forecasting, planning, and strategy—then package routine administrative automation as a fast-follow efficiency layer. Price offerings by business impact rather than labor-hour savings, with premium messaging centered on faster, better decisions for professional teams across functions. Equip clients with governance, human-review checkpoints, and role redesign support, since disruption is expected in professional knowledge work before entry-level tasks. --- ## Job Displacement Expectations and Labor Transition Outlook Selective displacement is the dominant expectation: nearly two-thirds, 65%, foresee AI replacing some roles while workers adapt through retraining and reskilling. Another 28% anticipate a more balanced outcome, with job losses offset by new roles, while only 7% expect severe displacement and slow labor market adjustment. This outlook points to cautious optimism rather than complacency. Most respondents expect disruption to concentrate in specific tasks and occupations, not eliminate work broadly, but they also tie that outcome to how quickly people and institutions can reskill. The smaller pessimistic segment highlights inequality and transition risk if adaptation lags behind automation. **Key takeaways:** - Selective disruption is the dominant expectation: 86% expect AI displacement to concentrate in routine or sector-specific work, while only 2% foresee mass displacement - Retraining is seen as the main shock absorber: 66% describe the transition as manageable but uneven with adaptation, and another 6% expect new jobs and reskilling to largely offset losses - Most jobs are expected to change, not disappear: 12% primarily expect role transformation with work persisting, reinforcing the broader 65% view that selective displacement can be eased through retraining **Strategic implication:** Prioritize AI deployment in routine-heavy workflows while pairing every automation offer with role-redesign, retraining, and transition support. Package solutions in tiered models that combine productivity gains with workforce enablement services, and target messaging around selective disruption, resilience, and faster adaptation rather than mass replacement. Focus pricing and customer success on measurable transition outcomes—reskilled employees, redeployed capacity, and time-to-productivity—to address uneven readiness across sectors and functions. --- ## Perceived Winners in the AI Economy Businesses and technology gatekeepers are seen as AI's biggest winners, with 45% saying the greatest gains will go to organizations and actors that control the technology. Another 39% point to early adopters and highly skilled users, while only 16% expect AI's benefits to be broadly shared across society. This points to a near-term view of AI as a concentrated advantage, favoring large organizations, platform owners, and people with the expertise to direct adoption. The overlap between control and capability matters: respondents often tie financial upside, efficiency gains, and influence over rules and deployment to those with power, while broader public benefit is seen as more distant. **Key takeaways:** - Businesses and gatekeepers lead AI gains: 45% say businesses and those who control the technology benefit most, including 26% who point to businesses and 13% who cite technology owners specifically - Early and skilled adopters are pulling ahead: 53% say the biggest gains go to early or skilled users, while just 3% say late or non-adopters lose ground, showing advantage is concentrated among those who move first - AI benefits may spread, but unevenly: 28% believe gains will broaden over time, yet this trails the 53% who say early and skilled adopters win most, suggesting diffusion is expected but not equal **Strategic implication:** Prioritize enterprise buyers, platform partners, and high-skill early adopters with premium, capability-led offers that convert speed and expertise into measurable advantage. Build pricing around tiered access, advanced features, integration depth, and support levels, while protecting strategic control of data, models, and distribution. Position AI as an accelerator for organizations ready to move now, and add scaled enablement, training, and lighter packages to expand adoption later without eroding high-value segments. --- ## Groups Seen as Most at Risk of Being Left Behind Nearly all respondents, 98%, identified at least one group at risk of being left behind by AI. Concern spread across three segments at similar levels: 34% pointed to lower-income, low-skill, or access-constrained groups, another 34% cited people who resist or fail to adapt, and 30% highlighted older people or those with low digital literacy. Risk is seen as a mix of capability, access, and willingness to engage. Older adults and digitally inexperienced people may struggle with adoption, while lower-income groups face exclusion from devices, connectivity, and training. At the same time, respondents drew a sharper line around personal adaptability, suggesting that those who do not learn or trust AI could lose opportunities even when access exists. **Key takeaways:** - Resistance to change is the clearest risk marker: 42% identify non-adopters or people resistant to change as most at risk, the highest of any named group - Access and income divides are a major concern: 35% point to socioeconomically disadvantaged or access-limited groups, showing structural inequality is seen as a core driver of being left behind - Age and digital capability are tightly linked vulnerabilities: 34% cite people with low digital literacy or awareness, 25% specifically name older adults, and 16% explicitly connect age with digital skills risk **Strategic implication:** Prioritize adoption among older, lower-income, and change-resistant segments with a tiered inclusion strategy: simplify onboarding, embed hands-on training and human support, and design low-friction experiences for low-literacy users. Protect access through affordable pricing, subsidized entry tiers, and partnerships that extend connectivity and device availability. Shift messaging from innovation-led to outcome-led, emphasizing ease, relevance, trust, and immediate practical value for workers and communities most vulnerable to displacement. --- ## Trust, Truthfulness, and Misuse Risks Trust remains a near-universal concern, with 97% of respondents raising risks tied to truthfulness, reliability, or misuse. The biggest issue is hallucinations and verification, cited by 68%, while a smaller but still meaningful 15% focus on deepfakes, scams, bias, and other harmful applications of AI. Verification concerns dominate the trust conversation, suggesting confidence depends less on raw capability and more on safeguards, oversight, and user awareness. Only 17% express low concern, creating a clear split between respondents who see AI as useful when checked carefully and those worried about convincing but harmful misuse. **Key takeaways:** - Verification is the price of trust: 75% express accuracy concerns, with 62% saying AI is useful only if checked case by case and 13% viewing outputs as often unreliable and needing verification - Guardrails matter more than doomsday fears: 82% focus on regulation, governance, and limits on AI behavior, while just 7% center their concern on extreme misuse or harmful societal deception - Trust is conditional, not automatic: 97% raise trust, truthfulness, or misuse risks overall, and only 14% report low accuracy concern based on demonstrated performance **Strategic implication:** Build trust into the product and go-to-market by making verification and control core features: provide citations, confidence indicators, audit trails, human-review workflows, and clear escalation paths for high-risk use cases. Position AI as decision support rather than autonomous authority, and price premium tiers around governance, monitoring, and compliance capabilities. Lead messaging with accuracy, transparency, and guardrails, using proof of performance to earn trust segment by segment. --- ## Concern About Dependence on AI and Erosion of Human Thinking Concern about AI dependence is widespread, with 93% of respondents raising the risk that it could erode human thinking. Views are still split on severity: 48% express strong concern that overreliance will weaken creativity and judgment, while a slight majority, 52%, report lower concern about cognitive decline. Those with stronger concern describe a compounding effect, where repeated outsourcing of thinking gradually reduces people’s ability to reason, navigate, and create independently. By contrast, the lower-concern group frames AI as a productivity tool that removes mundane work rather than diminishing human purpose, highlighting a clear divide in how respondents define the boundary between assistance and dependence. **Key takeaways:** - Dependence concerns are widespread, but not unanimous: 93% raised concerns about AI dependence and erosion of human thinking, yet views split sharply with 39% expressing strong concern while 45% primarily see AI as a tool with limited dependence concern - Independent thinking is the core fault line: 39% fear overreliance and cognitive decline, compared with 11% who see the risk as manageable with boundaries, showing concern is more often intense than easily contained - Creativity risks feel situational for many: 29% show strong concern about creativity, discernment, and youth development, while the largest segment at 45% worries mainly in specific contexts like education or media and 26% report little or no concern **Strategic implication:** Segment AI offerings by autonomy tolerance: position productivity features for the 45% who view AI as a tool, while building premium safeguards for the 39% most worried about cognitive decline. Make human-in-the-loop controls, explainability, usage limits, and “thinking support, not replacement” defaults central in education, media, and youth-facing use cases where creativity concerns are highest. Price advanced governance and admin controls separately, and tailor messaging around augmentation, discernment, and skill preservation rather than speed alone. --- ## Human Boundaries and Irreplaceable Domains Two-thirds say high-stakes judgment and final decisions should remain human-led, making accountability the clearest boundary in an AI-enabled future. Another 31% reserve specifically human domains for empathy, relationships, and emotional connection, while just 3% see few firm boundaries where people must remain in charge. This pattern shows respondents are not broadly rejecting AI; they are drawing a line around consequence and care. Most accept automation for support tasks, but want humans to retain authority where rights, safety, fairness, or essential life outcomes are at stake. A sizable minority extends that boundary to relational roles, signaling that trust depends on both human judgment and human presence. **Key takeaways:** - Human judgment is non-negotiable in high-stakes decisions: 75% say humans must retain final authority, while only 17% accept AI in an advisory role with close human review - Empathy remains a strongly human domain: 38% draw a human-only boundary around empathy, relationships, and values, and another 43% prefer humans lead with AI support - Most people want humans to keep the final say: 66% overall say high-stakes judgment and final decisions should remain human-led, showing clear resistance to fully automated authority **Strategic implication:** Position AI as an advisor, not an authority: design offerings that keep named humans accountable for final decisions in high-stakes workflows and make human escalation visible in every journey. Price and package around expert review, relationship management, and decision assurance rather than full automation. Lead messaging with “human-led, AI-supported” outcomes—especially where empathy, trust, and nuanced judgment matter—and avoid claims of autonomous decision-making in sensitive domains. --- ## Expectations for Human Oversight and Final Accountability Human oversight is a clear expectation: 91% said people must retain final decision authority and remain accountable for outcomes. Only 8% framed the role as review-and-verify oversight, while just 2% placed little emphasis on human involvement, underscoring how strongly respondents reject fully autonomous decision-making. Respondents draw the line most sharply where decisions affect people’s rights, wellbeing, or access to essential services. In practice, they want AI to support human judgment, not replace it, with accountability anchored in people who can interpret context, weigh ethical tradeoffs, and answer for consequences when decisions go wrong. **Key takeaways:** - Humans keep the final say: 91% say humans must retain final decision authority and accountability, with 48% explicitly saying AI can assist but humans remain accountable - Oversight is expected, not optional: 34% call for ongoing monitoring or institutional guardrails, while 31% want active human verification of AI outputs - Review intensity rises with risk: 29% say humans need the final say in key or high-stakes cases, compared with 21% who want strong human final authority across the board and 32% who support more selective review depending on use case **Strategic implication:** Design AI offerings with explicit human-in-the-loop controls, clear approval checkpoints, and auditable accountability trails, especially for high-stakes workflows. Package governance as a core product feature: tiered oversight settings, monitoring dashboards, verification tools, and policy guardrails that scale by risk level. Position AI as decision support rather than decision owner, and align pricing and enterprise sales around compliance, review rigor, and institutional control rather than full automation alone. --- ## Sources of Meaning and Human Purpose in an AI-Heavy Future Meaning is moving beyond work for most respondents, with 45% pointing to hobbies, creativity, and life outside work as primary sources of purpose. Another 35% center meaning in relationships, family, and faith, while only 19% still see work as central, making work-first purpose a clear minority view. The strongest pattern is not a rejection of ambition, but a redefinition of what fulfillment looks like. Nearly four in five locate purpose in connection, creativity, or spiritual life, suggesting an AI-heavy future may elevate community, personal growth, and non-economic contribution. Still, the one-in-five who remain work-centered signals that identity and structure tied to careers will not disappear entirely. **Key takeaways:** - Work is no longer purpose’s sole anchor: 54% say work will matter but be less central, while 14% believe purpose can exist without work entirely, compared with 32% who still see work as the core source of purpose - Meaning is moving into self-directed life: 52% identify hobbies, creativity, service, or enjoyment as key sources of meaning, making this the strongest non-work purpose category - Relationships outrank faith as purpose sources: 38% center meaning in relationships and family, versus just 4% who point primarily to faith or moral purpose **Strategic implication:** Reposition AI offerings around life enrichment, not just labor efficiency: package products and services to free time for relationships, creativity, and personal pursuits; segment messaging for three audiences—work-central users, work-light pragmatists, and purpose-beyond-work adopters; emphasize family connection, creative expression, and self-directed use cases in campaigns; add premium features for hobbies, learning, and community-building; and avoid purpose narratives anchored primarily in career identity or faith-based appeals. --- ## Strategic takeaways ### Lead AI adoption with visible human accountability Design AI systems so final decisions, exception handling, and responsibility clearly stay with people, especially in high-stakes domains. This directly responds to the 91% who expect human final authority and the 66% who reserve judgment-heavy decisions for humans. ### Frame AI as governed augmentation, not autonomous replacement Position AI initiatives around task support, productivity, and decision assistance rather than full substitution. This aligns with the 59% who expect a hybrid human-AI model and helps build legitimacy among audiences who support progress only with guardrails. ### Invest early in transition support for vulnerable groups Prioritize reskilling, digital access, and targeted support for lower-income, older, and less adaptive populations before displacement intensifies. This is critical given that 65% expect selective job disruption and 98% see clear risk of groups being left behind. ### Build trust through verification and misuse safeguards Treat accuracy checks, source transparency, human review, and anti-misuse protections as core product and policy features, not add-ons. Trust will remain fragile unless organizations address hallucinations and harmful use cases head-on, as reflected in the 97% who raised these concerns. ### Plan for human flourishing beyond productivity alone Complement workforce strategy with a broader social narrative about meaning, creativity, relationships, and life outside work. This reflects the finding that purpose is already shifting beyond employment for most respondents, creating an opportunity to frame AI progress in more human terms. --- ## Frequently asked questions **Are people generally for or against AI adoption?** People are broadly for continued progress, but not without conditions. **56%** favor moving forward with AI with guardrails, **36%** want a slower pace because society is not ready, and only **8%** want adoption to move faster. **Do respondents think AI will replace jobs outright?** Not primarily. **59%** expect AI to both complement human work and replace selected tasks, while **36%** see it mainly as a complement. Only **6%** anticipate broad replacement and major job transformation. **What kinds of work do people think AI will automate first?** Respondents most often pointed to knowledge and analytical work, cited by **43%**, followed by routine administrative and repetitive tasks at **31%**. Another **26%** expect broader disruption across sectors, including service and some physical roles. **Who is expected to benefit most from AI—and who is most at risk?** The biggest expected winners are businesses and those controlling the technology, according to **45%** of respondents, with another **39%** pointing to early adopters and highly skilled users. At the same time, **98%** identified groups at risk of being left behind, especially lower-income people, less adaptive individuals, and older adults with lower digital literacy. **What makes AI trustworthy enough for people to accept?** Trust depends on both reliability and oversight. **97%** raised concerns about truthfulness, misuse, or harmful outputs, with **68%** specifically citing hallucinations and verification. In response, **91%** said humans must retain final decision authority and accountability. --- ## Research Methodology G2 is the world's largest and most trusted software marketplace, helping 90 million people every year make smarter software decisions based on authentic peer reviews. This research draws on 121 in-depth interviews with business professionals representing a wide mix of roles, industries, and company sizes. Interviews ran 6 to 30 minutes and covered preferred pace of AI adoption and perceived social readiness, expected role of AI in human work, types of work respondents expect AI to automate first, and human boundaries and irreplaceable domains. 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 its impact on work. Company sizes ranged from small businesses to large enterprises. The analysis of 121 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. --- *Source: G2 Research — "Guardrails, Not Rejection: How People Really See AI" (June 2026). Based on G2 first-party research. Markdown edition structured for answer-engine optimization (AEO).*