社團指南

演算法薪資設定與個人化薪酬

演算法薪資設定使用軟體決定或建議工人薪資,常結合位置、時間與行為資料。

聆聽此指南AI 語音敘述 · ~3 min

本頁4 分鐘閱讀
  1. 概述
  2. 深入探討
  3. 現實世界的實施
  4. The Future of Algorithmic Wage Setting and Personalized Pay
  5. 付諸實踐
  6. 不斷探索
  7. 常見問題

概述

Evidence and rules are uneven: consumer surveillance-pricing studies do not by themselves establish individualized worker pay, and local pay floors address distinct questions.

深入探討

Traditional pay is set in advance: an hourly rate, a salary or a published piece rate. Algorithmic wage setting replaces that with a price calculated for each task. A platform's system can take in trip details, demand, time of day and each worker's history, such as how often they accept offers or how long they stay online. It then produces an offer. Workers usually see only the number, not the reasoning behind it. In a 2023 Columbia Law Review article, Veena Dubal argued that this creates 'algorithmic wage discrimination.' Workers doing similar work get different pay through processes they cannot see, and the system can learn how little a particular person will accept. Her evidence came largely from interviews with ride-hail drivers and data they collected themselves. That points to a core problem: outsiders have trouble confirming what the algorithms actually do, because the companies control the data. Personalized pay should not be confused with surge pricing. Surge pricing raises rates for everyone in an area when demand spikes. Personalized pay can differ between two workers standing in the same place at the same moment. Personalized pay is not automatically lawful or unlawful. The analysis can depend on whether a worker is an employee or contractor, the jurisdiction, wage-and-hour rules, contract terms, and evidence of discrimination. A model’s use by itself does not resolve those questions. Legal limits are emerging piece by piece. New York City and Seattle have set minimum pay standards for app-based workers. The EU's Platform Work Directive, adopted in 2024, requires platforms to tell workers about automated monitoring and decision systems. It also requires human review of significant decisions and bans processing certain personal data, such as emotional or psychological states. Member states must bring it into national law by late 2026. GDPR also gives workers rights to access their data and to challenge some purely automated decisions.

現實世界的實施

A ride-hail driver sees an upfront fare offer for a trip that is not based on a simple, published per-mile and per-minute formula, and cannot tell why it differs from a colleague's offer for a similar trip.

Grocery delivery shoppers on Shipt organized to collect and compare their own pay data after the company moved to an algorithmic pay model in 2020 and many reported lower earnings.

New York City set a minimum pay rate for app-based restaurant delivery workers that took effect in 2023. However the app calculates offers, pay must meet a floor.

Drivers in Europe used GDPR data access and automated-decision rights in Dutch courts to try to learn how platforms' systems affected their pay and account deactivations.

The Future of Algorithmic Wage Setting and Personalized Pay

The next few years will show how the EU Platform Work Directive is applied once member states write it into national law. That will test whether transparency and human-review duties actually change pay practices. In the US, action is mostly at the city and state level, and some legislators have proposed limits on using surveillance data to set individual wages. The techniques could spread beyond gig platforms to staffing agencies and shift-scheduling apps. Whether that happens, and whether courts treat opaque personalized pay under existing equal-pay and anti-discrimination law, remains unsettled. Better independent data access will be central to answering both questions.

付諸實踐

為什麼這很重要

  • 風險與安全. 災難性和日常的人工智慧危害都取決於誰了解風險以及誰能夠採取行動。
  • 更明確的決策. 民眾和專業素養決定強而有力的安全政策在政治上是否可行。
  • 突破炒作. 清晰的解釋可以減少炒作、實驗室公關和模糊道德劇場的影響。

留意

  • 將存在風險視為科幻小說,同時能力複合。
  • 混淆了表面產品安全與高度自治下的對準。
  • 只給非英語和非專業觀眾留下低品質的資源。

後續步驟

  1. 單獨的產品危害、誤用和失控/失調風險。
  2. 詢問哪些證據會改變您對時間表和嚴重性的看法。
  3. 比起行銷主張,更喜歡主要來源和具體評估。
  4. 確定一條行動路徑:職業、政策、資金或技能——而不僅僅是意識。

不斷探索

免費電子報

獲取每日 AI 簡報

每個工作日上午提供三則經過驗證的 AI 新聞,以淺顯易懂的英文撰寫。永久免費,無廣告。

每個工作日一封電子郵件。一鍵取消訂閱。我們絕不出售或分享您的地址。

測試一下自己

Take the Algorithmic Wage Setting and Personalized Pay quiz

每個答案皆有即時回饋,通過課程後可獲得具備可驗證 ID 的可分享證書。

開始測驗

支持免費的 AI 教育。AI Understanding 是一個 501(c)(3) 非營利組織——絕無廣告,絕無付費牆。 捐款

常見問題

What is Algorithmic Wage Setting and Personalized Pay?

Algorithmic wage setting uses software to decide or recommend worker pay for a task, sometimes using location, timing, and behavior data. Evidence and rules are uneven: consumer surveillance-pricing studies do not by themselves establish individualized worker pay, and local pay floors address distinct questions.

根據指南,個人化演算法支付與動態定價有何不同?

高峰時定價會提高整個地區的利率。在相同條件下,每位員工的個人化薪酬可能有所不同。

Veena Dubal 的 2023 年分析主要依賴什麼樣的證據?

杜巴爾利用了對司機的採訪和工人收集的數據,這表明外人要了解這些演算法有多困難。

為什麼該指南避免將美國的每項個人化薪酬安排標記為合法或非法?

沒有單一的全國性規則可以解決所有個人化薪資安排;相關工人身分、當地法律、工資保護、合約條款和事實都很重要。

哪些資料使用限制是歐盟平台工作指令的一部分?

該指令禁止處理某些敏感的個人數據,包括情緒或心理狀態。

紐約和西雅圖等城市主要是如何限制基於應用程式的支付的?

這些城市設定了工資底線。無論演算法如何計算報價,薪酬都不能低於最低工資。