Does delegating work to AI actually damage worker skills?

Anthropic's June 2026 Economic Index report states that "the people who delegate to Claude the most are the most optimistic about their future labor market outcomes, and feel their skills are growing in value." The report flags this as counter to "a common concern" that heavy reliance on AI erodes a worker's own capability rather than building it. It also notes that expectations about AI's general trajectory are "strikingly uniform" across respondents regardless of experience, geography, or job exposure — people broadly expect "significant AI progress over the next year" — but views on what that progress means for the respondent's own career are "less uniform." Early-career workers, specifically, "report that AI can do the highest share of their work and express the most concern about job loss."

The report offers no causal mechanism for the delegation-optimism link, only the correlational framing above. Its account of why this might make sense is aspirational rather than mechanistic: respondents' "hopes for the next decade center not on replacement but on collaboration," wanting AI to "preserve meaningful work and automate the drudgery" with gains "shared widely." The excerpt does not test, and does not claim to rule out, the reverse direction — that workers who are already confident about their prospects are the ones willing to delegate more, rather than delegation itself producing the confidence.

This sits in tension with Does AI assistance erode the skills needed to oversee it?, where a different Anthropic-adjacent survey finds engineers worried that delegating to Claude wears down the oversight skills needed to supervise it. One measures self-rated skill erosion among people who must verify AI output; the other measures self-rated career optimism among people who delegate work away — both from respondent pools close to Anthropic's own product. It is also worth reading against Which workplace cues survive AI mediation and which disappear?: if effort and skill-building genuinely recede into AI-mediated output for heavy delegators, a reported feeling that "skills are growing in value" may reflect confidence in one's own judgment and oversight role rather than actual growth in task-level skill.

The excerpt gives no sample size, no definition of "delegate the most," and no independent measure of skill or earnings to check the self-report against — this is Anthropic surveying its own users about its own product, a population plausibly self-selected toward enthusiasts. The correlation between delegation intensity and optimism does not establish that delegation causes the optimism or that the optimism is well-founded; it is equally consistent with confident workers choosing to delegate more. Read narrowly, the finding says only that within this surveyed population, heavy delegation and career optimism travel together — not that AI delegation is safe for skills or employment more broadly.