概述
Automated screening can affect access to interviews, so employers should examine how tools are used, whether selection has adverse impact, and which notices or accommodations apply.
深入探討
Applicant tracking systems manage applications; some products include automated features, but not every ATS ranks candidates. Under Title VII, selection procedures can raise discrimination concerns when they have adverse impact. EEOC guidance addresses assessing impact in software and AI and explains job-related validation and business necessity. The four-fifths rule is a rule of thumb, not a guarantee of compliance. NYC Local Law 144 applies to covered automated employment decision tools and requires an audit within one year, public audit information, and notices; DCWP states notice timing. Disability law may also require reasonable accommodation in AI-assisted hiring. NYC Local Law 144 requires covered employers to conduct an annual independent bias audit, publish a summary, and give candidates at least 10 business days’ notice before using an AEDT; the notice must include job qualifications and characteristics used. The four-fifths rule compares a group’s selection rate with the highest group rate and is a rule of thumb, not a safe harbor. EEOC testimony also states employers using vendor screens remain responsible for understanding the product and determining whether it causes adverse impact.
現實世界的實施
An ATS stores applications and may include parsing features; behavior depends on product and configuration.
An employer examines selection rates and potential adverse impact for an automated screen.
A New York City employer checks whether Local Law 144 covers its tool and follows audit and notice rules.
An applicant who needs an accommodation asks the employer about an alternative process.
人工智慧履歷篩選的未來
Requirements vary by jurisdiction and can change. Check current rules, document tool use, provide required notices, and evaluate outcomes rather than assuming one audit makes a system fair or lawful.
付諸實踐
為什麼這很重要
- 配裝選擇. 應用級設計決定了人工智慧是否能改善實際結果。
- 團隊與工作流程. 良好的工作流程整合可以創造使用者值得信賴的生產力效益。
- 風險與安全. 範圍明確的用例可以減少變更疲勞和實施風險。
留意
- 將損壞的流程自動化可能會加劇現有問題。
- 團隊可能會過度自動化並消除所需的人工判斷。
- 如果不持續評估輸出,品質可能會出現偏差。
後續步驟
- 繪製目前工作流程並確定摩擦最大的步驟。
- 在完全自動化之前定義人工檢查點。
- 對使用者進行提示、升級路徑和品質標準的訓練。
- 追蹤任務級結果以確認持續價值。
不斷探索
常見問題
什麼是 AI 履歷篩選?
Employers may use software to organize, parse, or evaluate applications. Automated screening can affect access to interviews, so employers should examine how tools are used, whether selection has adverse impact, and which notices or accommodations apply.
什麼樣的軟體通常具有人工智慧履歷篩選功能?
履歷篩選通常內建於申請人追蹤系統中,該系統是雇主用來收集和管理工作申請的平台。
當招募選拔程序根據第七章產生不利影響時,哪些法律原則是相關的?
平等就業機會委員會指南解決了與工作相關的驗證和業務必要性,其中選擇程序會產生不利影響。
紐約市地方法 144 對承保範圍內的 AEDT 使用有何要求?
DCWP 列出了所涵蓋工具的審核、公共資訊和通知要求。
為什麼從履歷中刪除名字仍然無法防止偏見?
其他欄位可以與受保護的屬性相關聯,因此即使隱藏名稱,模型仍然可以產生偏差的結果。
常用的五分之四法則是什麼?
平等就業機會委員會將此稱為選擇率差異很大的經驗法則,而不是安全港。