If your organisational communication strategy on AI's impact on work is calibrated against expert or industry framing (that AI will augment human capability, transform jobs positively, expand the economy), the 2024-2025 Pew survey of US adults shows the framing is misaligned with majority public sentiment by 50 percentage points on the core question. 73% of AI experts said AI will have a positive impact on how people do their jobs over the next 20 years. Only 23% of US adults agreed. The same divergence shows up across nearly every domain Pew tested: the economy (69% experts positive vs 21% public), medical care (84% vs 44%), K-12 education (61% vs 24%). When experts and the public disagree this systematically, organisational communication strategy needs to absorb both signals rather than pick one.

US perceptions of AI's societal impact: public vs experts, 2025

The 50-point gap on jobs is the largest divergence in the dataset and the most consequential for workforce planning. Three structural observations support this reading.

The first observation: the gap holds consistently across domains. Where AI is plausibly transformative (medical care, education, the economy, jobs), experts are systematically more positive than the public. Where AI intersects with trust and social connection (elections, news, personal relationships), both groups are negative, with experts only modestly more positive. The pattern suggests the divergence is not about whether AI works (both groups agree it works in technical-capability domains) but about whether AI's effects on people will be net positive.

The second observation: the public's pessimism is grounded in concrete labour-market expectations. Per the same Pew survey, 64% of US adults expect AI to lead to fewer jobs in the next 20 years; only 5% expect more jobs. Experts are less pessimistic (39% predict fewer jobs, 19% more jobs) but their distribution still leans negative. The public's pessimism is more extreme but directionally consistent with the expert view. Both groups expect AI to reduce employment; they disagree on magnitude and on whether the people doing the remaining jobs will be better or worse off.

The third observation: the 2035 outlook survey from the Elon University Imagining the Digital Future Center shows the same pattern on human capacities. US adults are more likely than AI experts to expect AI to have negative effects on metacognition (53% adults vs 36% experts), decision-making (48% vs 30%), and social and emotional intelligence (51% vs 34%). Mental well-being is the only category where the two groups are close (55% adults, 53% experts). The pattern: the public expects AI to degrade specifically human capabilities that work depends on; experts are less concerned, except on mental well-being where both groups converge.

Taken together, the systematic divergence between expert and public sentiment on AI in work and society is the largest signal in the public-opinion data. It shapes how AI deployment is received by employees, customers, voters, and policymakers, none of whom share the expert framing as the default.

Three structural implications follow for organisations whose communication, deployment, and change-management strategy depends on how AI is received.

The first implication: AI deployment communication strategies that rely primarily on expert framing will under-resonate with the majority of employees, customers, and citizens. The "AI will augment your work, expand the economy, improve outcomes" framing is the expert position. The public position is closer to "AI will displace jobs, change work in ways I can't control, and may degrade the human capacities I rely on." Communication that doesn't engage with the public framing will be received as not engaging with the actual concerns.

The second implication: medical care is a particular case. The 84%-44% gap (experts vs public) on medical AI is one of the largest in the dataset, but the public position is also notably less negative than for jobs. The 44% positive on medical AI suggests the public sees specific use cases (diagnostics, drug discovery) as plausibly beneficial even where general AI deployment is viewed with concern. Healthcare-adjacent organisations have a structurally different communication challenge than employment-adjacent organisations: the public starts at a less negative baseline and is more open to specific positive use cases.

The third implication: the divergence on personal relationships, elections, and news (where both groups are negative) is the lowest-resonance ground for any AI communication. Organisations deploying AI in these adjacent domains (political communication, news media, social connection products) face scepticism from both expert and public audiences. The communication strategy in these domains differs from the strategy in employment or healthcare; reassurance and capability framing both face structural scepticism.

The forecasting data confirms the divergence is structurally large and persistent across surveys, domains, and methodologies. The 2025 LEAP forecasting research (Forecasting Research Institute) confirms the pattern at the forecasting level: across 68 AI capability forecasts, public views align with experts in only 9% of cases. When they diverge, the public expects slower progress 71% of the time. The pattern is robust and not a survey artefact.

For organisations setting AI deployment, change-management, and communication strategy in 2026, planning needs to engage with the public framing as the dominant reception environment. The expert framing is the technically accurate view; the public framing is what employees, customers, and citizens will be reading AI deployment through. Strategic communication that bridges the gap, acknowledging the public concerns while making the expert case for specific deployments, produces better reception than strategy anchored on either framing alone.

The 50-point expert-public gap on AI helping jobs, the 64% public expectation of fewer jobs, and the consistent direction of public pessimism on AI-affected domains together describe an organisational communication challenge that 2026 strategic plans need to absorb. The plans that read the public sentiment accurately will outperform plans that operate from the expert framing alone. The plans that engage with both perspectives, acknowledging the legitimate concerns the public sentiment is grounded in while making evidence-based cases for specific AI deployments, will be most effective at building trust through AI rollout.