# AI-Use Visibility Is a Proxy Screen Not a Competence Signal ## Claim Screening candidates for visible AI-tool adoption selects for the appearance of adopting a trend, not for the capability a technical role actually needs — the same failure mode as any other convenient proxy credential. <!-- atomicity: 1 idea ✓ --> ## Scope & conditions - Holds where the actual job requirement is verifying, debugging, or extending code correctly — capabilities AI-tool enthusiasm doesn't reliably predict. - Doesn't claim AI fluency is worthless — it can be a legitimate, task-specific requirement when the role genuinely centers on AI-assisted workflows and that's stated as such. - Weakens when a hiring rubric ties AI use to a named task and evaluates the output (does the candidate catch outdated or unsafe generated code?) rather than treating tool adoption itself as the qualification. ## Evidence Andrew Norcross reports a prospective employer rejected him because he "did not use enough AI" in his own workflow — a judgment about tool visibility, not about whether he could do the job. He separately notes that AI is "pretty useful for mundane stuff... boilerplate" but resists being reduced to reviewing generated output, since he sees that as discarding the expertise he built. The market correction he anticipates runs the other direction too: "These people that know AI don't actually know how to do the job" — meaning tool fluency fails as a proxy in both directions, for and against actual competence. ## Links - [[3. Resources/Permanent Notes/Credential Screens Select for the Proxy Not the Work]] — because extends; this is the same proxy-screening failure (there: algebra tests for mechanical work; here: AI-tool visibility for coding competence) applied to a new domain. - [[3. Resources/Literature Notes/Don't Use AI You Might Not Get the Job]] — because evidence-for; source literature note with full timestamped testimony. - [[3. Resources/MOCs/AI Cost & Tokenomics MOC]] — because applies-in; hiring-signal distortion is a labor-market cost of the current AI adoption pressure. - [[3. Resources/MOCs/AI & the Future of Knowledge Work MOC]] — applies-in: Joins AI and hiring or employment notes in the knowledge work cluster. ## Tests / falsifiers - Compare job performance (defect rate, code review outcomes, time-to-resolution) between candidates screened on visible AI-tool use versus candidates screened on task-specific work samples — if AI-use screening predicts performance as well as work samples, this claim weakens. - Find a hiring rubric that ties AI use to a specific task and evaluates verification quality rather than raw adoption — if that's already standard practice in WordPress hiring, the "proxy, not signal" framing may already be resolved in parts of the market. ## Maintenance - Last tended: 2026-08-16 - Next review: 2026-11-16 ## Source - [[3. Resources/Literature Notes/Don't Use AI You Might Not Get the Job]] — The WP Minute, Eric Karkovack interviewing Andrew Norcross.