AI-use image concerns

AI-use image concerns are reputational or evaluative fears that visible AI reliance will signal something undesirable about the worker: weak judgment, low skill, low effort, replaceability, or poor professional identity. The construct matters because it creates an adoption barrier even when the tool is accurate, available, and economically useful.

2026-almog-barriers-ai-adoption gives the cleanest behavioral evidence so far. In a field experiment with 450 U.S.-based Upwork workers completing an AI-assisted image-categorization task, making AI reliance visible to an HR evaluator reduced workers’ AI recommendation following by 4.3 percentage points overall and 8.0 percentage points on disagreement rounds, lowering final accuracy from 79.1% to 76.4%. The design is important: evaluators were explicitly instructed to assess expected accuracy on the same AI-assisted task, so the adoption penalty persisted even when the normative criterion should have favored effective AI use.

The mechanism is not simply generic AI distrust. Almog’s platform-feedback elicitation found that when feedback would disclose above- or below-average AI use, workers more than doubled their selection of confidence in judgment as the trait to emphasize. In a second-stage evaluator task, workers also penalized AI reliance in others’ profiles: an additional percentage point of AI reliance reduced evaluation scores by 0.36 points, while an additional point of accuracy increased scores by about one point. This makes image concerns socially self-reinforcing: workers avoid visible AI use partly because they themselves read visible AI use as a negative signal.

2025-mayer-generation-ai-job-crafting supplies a complementary qualitative mechanism among entry-level consultants. As managers struggled to distinguish individual knowledge and performance amid widespread GenAI use, some professionals engaged in signal crafting: displaying their GenAI capability, work process, or diligence to affirm that their contribution remained distinct and valuable. This is not direct evidence of the experimentally observed penalty in 2026-almog-barriers-ai-adoption; it suggests that contested professional legibility can prompt active self-presentation as well as avoidance of visible reliance.

2026-ghosh-new-social-image broadens the mechanism upstream of visible use. In a small video-vignette experiment, a more competent or proactive AI coworker generally reduced favorable self- and peer-perceptions of ownership, meaningfulness, satisfaction, role dynamics, and affect. This is not evidence that workers will conceal actual AI use, but it suggests that the image problem can begin when a system’s initiative or capability makes the worker’s own contribution feel less socially legible.

Connections

• Relates to ai-adoption because adoption can be blocked by social meaning and evaluation risk, not only by tool access, skills, readiness, or use-case discovery.
• Relates to task-level-ai-adoption because the same task-level opportunity can invite or suppress AI use depending on whether reliance is private, visible to evaluators, or attached to worker reputation.
• Relates to human-ai-collaboration because collaboration quality depends on whether workers can follow useful AI advice without feeling that delegation undermines their standing as competent agents.
• Relates to ai-receptivity because willingness to use AI is partly status- and identity-contingent; workers may be technically receptive in private but behaviorally unreceptive under evaluation.
• Relates to ai-mediated-work-experience because visible AI use can affect autonomy, competence, identity, and perceived job quality even before formal job redesign occurs.

Contradictions & open questions

• The current evidence is strong for short-cycle remote image categorization under evaluator visibility, but the wiki needs field evidence from richer professional work where AI use may signal adaptability, laziness, strategic judgment, or policy compliance depending on local norms.
• Visibility may cut both ways. Some organizations may reframe AI reliance as evidence of adaptability or productivity, while others may implicitly penalize it as weak independent judgment; the boundary conditions are not yet mapped.
• 2025-mayer-generation-ai-job-crafting indicates that workers may respond to ambiguous evaluation by making AI-mediated competence and process visible, but the single qualitative case does not establish whether signal crafting changes evaluator judgments, advancement, or actual work quality.
• Reducing individual visibility could improve adoption, but it may also reduce auditability, learning, and accountability. The design problem is not simply to hide AI use; it is to decide when visibility should attach to the work system, the team, the tool, or the individual worker.