Salesforce survey links U.S. AI skepticism to bad pilot experiences, not just fear of job loss

Salesforce says a key hurdle to enterprise AI adoption is not only concern about whether artificial intelligence will eliminate jobs, but also whether workers trust the day-to-day tools they are asked to use. In a survey of more than 1,500 desk workers across multiple countries, Salesforce found that American workers were 43% more likely than the global average to describe themselves as skeptical of AI, a gap the company frames as a practical blueprint for turning limited “pilots” into consistent workplace usage.

The survey asked respondents whether they saw themselves primarily as AI skeptics or advocates. Salesforce reported that 53% of people in the U.S., U.K., and France identified as skeptics, compared with 26% in Mexico. Salesforce also suggested that countries with higher skepticism tended to be somewhat less likely to treat AI as part of their core workflow, implying adoption is constrained by more than corporate messaging.

Salesforce’s analysis points to the quality of the employee experience during AI trials. American workers, the company said, reported that their AI pilots did not work out because the tools produced generic or “untrustworthy” outputs. Salesforce tied the problem to the underlying data and platform foundations used by enterprises, arguing that user trust depends on whether systems generate reliable, context-appropriate results.

The company also said the divide is about what happens after an initial failure, not the presence of failure itself. In Salesforce’s survey, workers with the highest adoption rates were more likely to report continuing to work through unsuccessful AI pilots before reaching consistent usage. The implication is that iteration, paired with institutional support, can convert early disappointments into eventual proficiency.

Salesforce highlighted a path to scale by describing a group it calls “AI’s A-team,” workers it said successfully graduated from AI pilots into consistent or core use. The company said it identified over 500 such workers and described four shared traits: training to improve how they use the technology, AI embedded directly into the tools they already rely on, secure and context-aware AI treated as “nonnegotiable,” and AI that can be tailored to their roles.

In that group, Salesforce reported outcomes that would matter to enterprise leaders focused on repeat usage. It said 76% of “AI’s A-team” became active AI advocates, and 63% used AI daily. Nathalie Scardino, Salesforce’s President and Chief People Officer, said in the release that getting “AI fluency” right changes AI from a technology novelty into a workforce advantage by moving from deploying tools to changing how work gets done.

Beyond Salesforce’s own findings, the release situates U.S. hesitation inside a wider trust gap captured by Stanford’s 2026 AI Index. In Stanford’s public opinion chapter, the United States showed the lowest trust in its own government to regulate AI responsibly among surveyed countries, at 31%. Stanford also reported a large split between AI experts and the U.S. public on whether AI will have a positive impact on jobs.

The release contrasts that with emerging-market perceptions of AI benefits, leaning on findings from KPMG’s research. Salesforce cited KPMG for a statistic that 90% of people in emerging economies expect to benefit from AI applications, describing AI there more as a route to upward mobility than a driver of job displacement. KPMG’s report also links this higher optimism to observed benefits and expectations of positive outcomes.

Still, there are limits to what can be concluded from the release alone. Salesforce does not provide the full question wording for every survey item in the article text, nor does it disclose how leaders’ internal practices were quantified. The story also does not publish measured adoption rates or governance controls at specific companies, meaning the cited relationship between skepticism and pilot experience should be treated as directional rather than causal. Separate findings like Stanford’s and KPMG’s are broad, and they do not explain the micro-level drivers of employee trust in particular tools.

For enterprise leaders, the immediate takeaway Salesforce emphasizes is operational: better outputs, better training, better embedding of AI into existing workflows, and stronger assurances around security and context awareness. The next thing to watch is whether companies can raise the share of workers who move from experimentation to daily use, and whether internal “AI fluency” programs reduce skepticism as quickly as they improve productivity in practice.