The AI Hiring Boom Need A Human Appeal Button.
On October 29, Randstad reported a sharp change in U.S. recruiting. In its survey, 42 percent of employers said job applications had surged, while 31 percent were moving away from public job boards toward curated talent networks. Employers understandably want to escape the flood of AI-generated résumés. Yet replacing an open application process with opaque screening creates a serious question: when a qualified applicant disappears from consideration because of bad data or a faulty assessment, who can put the decision right?
Consider someone returning to work after two years caring for a parent. Her résumé shows an employment gap, but it says little about the skills she kept current. Or consider a technician who earned industry certifications instead of a university degree. An automated filter could downgrade either candidate without recognizing the relevant experience. The immediate victims are job seekers; the organizations that delegate consequential judgments without meaningful review bear responsibility for correcting errors.
U.S. authorities have begun addressing the problem. Under New York City's Local Law 144, employers using covered automated employment decision tools must arrange annual bias audits, publish summaries, and notify affected candidates. Those safeguards matter, but a public audit cannot correct one applicant's inaccurate record. A notice that software was used offers limited protection if nobody can reconsider the result.
The stakes also reach beyond the United States. An International Labour Organization and World Bank study covering 135 countries warned that developing economies could experience AI-related job disruption before receiving productivity gains. Clerical and administrative roles have provided routes to better employment, especially for women and younger workers. Closing off access to opportunity through unchallengeable screening would deepen an already serious risk.
UNESCO's Recommendation on the Ethics of Artificial Intelligence gives governments a clear direction. It calls for accountable systems, human oversight, and mechanisms to remedy harm. That international commitment should now become a practical standard in employment: no applicant should have to accept a consequential automated rejection without a meaningful way to contest an error.
Employers should give applicants three rights. Tell them when automated tools substantially influence screening and what qualifications are assessed. Let them correct inaccurate information or supply context that the system missed. Then provide an appeal to a trained human reviewer who can inspect the evidence, explain the result, and reverse an erroneous decision. The process should work without an attorney, a fee, or specialized technical knowledge.
Put the appeal route directly in rejection notices, offer it in accessible languages, and commit to response deadlines. Record which decisions change and why, while checking whether particular groups struggle to obtain review. Give recruiters enough time to investigate errors instead of rewarding them solely for the number of applications processed. If an applicant's only appeal is another chatbot repeating the first answer, the safeguard has failed.
National labor authorities should establish minimum notice and correction standards. International organizations can compare appeal access and reversal rates across countries and share effective practices. Employers need not wait for a treaty to act. Hiring technology can make recruitment faster, but speed cannot become an excuse for leaving a person's livelihood at the mercy of an unaccountable score.
written by Gleb Tsipursky
Comments