行業指南

法院中的人工智慧和司法風險評估

Judicial risk assessment tools are statistical models that estimate how likely a defendant is to miss court or be rearrested.

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 AI in Courts and Judicial Risk Assessment
  5. 付諸實踐
  6. 不斷探索
  7. 常見問題

概述

Judges use the scores in bail, sentencing and parole decisions. The tools can make decisions more consistent and reduce unnecessary detention, but they raise hard questions about fairness, transparency and due process, as the debate over the COMPAS tool showed.

深入探討

Risk assessment in criminal justice is older than machine learning. Many tools are fairly simple: a points system or regression model built from historical data. They score factors such as age, prior convictions, prior failures to appear and pending charges. The Public Safety Assessment (PSA), developed with funding from Arnold Ventures, draws on age and criminal history without interviewing the defendant. It produces separate scores for failure to appear and for new criminal activity, plus a flag for new violent criminal activity. COMPAS, sold by Northpointe (now Equivant), is a proprietary tool used in pretrial, sentencing and corrections decisions in some places. It also uses questionnaire answers. In 2016 ProPublica analyzed COMPAS scores in Broward County, Florida. It reported that Black defendants who did not reoffend were nearly twice as likely as white defendants to have been labeled higher risk, and that white defendants who did reoffend were more often labeled lower risk. Northpointe replied that the scores were equally predictive for both groups: a given score matched similar reoffense rates regardless of race. Both claims can be true at once. Researchers including Jon Kleinberg and Alexandra Chouldechova showed that when base rates differ between groups, a score generally cannot be calibrated and also have equal false positive and false negative rates. Choosing a definition of fairness is a policy decision, not a purely technical one. There are other concerns. Rearrest stands in for reoffending, and it reflects where police patrol. Proprietary models make scores harder to challenge. Judges may defer to scores, or override them selectively. Supporters answer that unaided judicial intuition is also biased and less consistent. A common misconception is that these tools are advanced AI. Most are transparent statistical models using a handful of factors. Most other AI in courts is administrative: sorting e-filings, transcription, translation and scheduling.

現實世界的實施

A pretrial services officer runs the Public Safety Assessment, which scores failure to appear and new criminal activity from factors based on the defendant's age and criminal history. The officer gives the report to the judge before the first bail hearing.

In State v. Loomis (2016), the Wisconsin Supreme Court allowed a sentencing judge to consider a COMPAS score. It required written warnings about the tool's limits and said the score could not be the deciding factor.

A county court uses software to sort incoming e-filings by case type and flag incomplete ones. It also sends automated text reminders about hearing dates, which studies have linked to fewer missed appearances.

A state's pretrial agency revalidates its risk tool on local data and finds that its 'high risk' cutoff flags more people than intended. The agency adjusts the threshold through a public policy review.

The Future of AI in Courts and Judicial Risk Assessment

As bail reform debates continue, jurisdictions have adopted, revised or dropped pretrial tools. Evidence on whether the tools reduce detention without raising crime is mixed, and it depends heavily on how judges and policies use the scores. Likely directions include more public validation studies, transparent or open-source models, and limits on using algorithmic scores in sentencing. On generative AI, courts are issuing guidance for judges and clerks who use chatbots to draft and summarize. That guidance generally stresses confidentiality and keeps humans responsible for decisions.

付諸實踐

為什麼這很重要

  • 背景與規則. 產業背景決定了人工智慧創意能否與現實接觸。
  • 品質管控. 領域約束會影響可接受的錯誤率和監督模型。
  • 配裝選擇. 成功的部署使技術能力與第一線工作流程保持一致。

留意

  • 監理要求可能會使原本強大的原型失效。
  • 歷史資料可能會編碼損害特定社區的偏見。
  • 遺留系統可能會造成整合瓶頸和隱性成本。

後續步驟

  1. 讓領域專家參與從問題框架到評估的整個過程。
  2. 在啟動前設計審計追蹤和文件。
  3. 儘早驗證合規性和安全義務。
  4. 分階段推出,並有明確的停止和回滾標準。

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 AI in Courts and Judicial Risk Assessment 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 AI in Courts and Judicial Risk Assessment?

Judicial risk assessment tools are statistical models that estimate how likely a defendant is to miss court or be rearrested. Judges use the scores in bail, sentencing and parole decisions. The tools can make decisions more consistent and reduce unnecessary detention, but they raise hard questions about fairness, transparency and due process, as the debate over the COMPAS tool showed.

公共安全評估 (PSA) 會產生什麼結果?

PSA 使用基於年齡和犯罪史的因素,在沒有訪談的情況下產生單獨的審前評分。它在審判前評估風險。它沒有提到任何有關內疚的事情。

ProPublica 2016 年對布勞沃德縣 COMPAS 的分析報告了什麼?

ProPublica 專注於誤報:被貼上高風險標籤但沒有再犯的人。黑人被告的錯誤率要高得多。

Northpointe 對 ProPublica 的調查結果有何反應?

Northpointe 為該工具進行了辯護。相同的分數意味著每組的風險大致相同。這是與相同錯誤率不同的公平性定義。

為什麼風險評分通常無法校準並且同時在各組之間具有相同的假陽性率和假陰性率?

Kleinberg、Chouldechova 等人的研究表明,當基本費率不同時,這些公平標準會發生衝突,除非是一些微不足道的情況。因此,選擇優先考慮哪一個是一項政策決定。

威斯康辛州最高法院在「州訴盧米斯」案 (2016) 中做出了什麼裁決?

法院允許謹慎使用有限的使用,平衡潛在有用的工具和對專有模型的正當程序擔憂。