Court accepts Spirit's Predict review stopped at 85% recall

In April 2020 a US magistrate judge found that Spirit AeroSystems' Predict review of 322,000 documents, stopped at 85% recall, was reasonable and that more was not proportional.

  • 1 for an organization

    Spirit reported that its review identified 85% of the responsive documents, and the court held that the review met its discovery obligations.

Depth runs from 1, a convenience or time saved, to 5, a life or a livelihood preserved.

Status
Reported benefit.
Evidence

foundstudiedreportedclaimed

An independent court document supports the benefit.

Compared with

Any other form of review. Assumed without a measurement.

The system’s part
Not isolated. The court accepted Spirit's 85% recall figure and did not test the tool or compare it with another method. Its view that other review would have produced fewer documents rests on earlier cases. The orders describe a tool that learns continuously from the coding of reviewers.
Later
Taken to court, 18 June 2020. On 18 June 2020 the court granted Spirit's motion to shift its TAR expenses, reported at about $600,000, to the plaintiff. It found that review of a largely non-responsive data set had become disproportionate to the needs of the case.
Not yet checked
LimitsCosts and harmsLearns from data

Download data Cite Suggest a correction

Benefit type
Improved Task Accuracy
System
Predict, a technology-assisted review tool that uses continuous active learning
Made by
Not named in the orders, which say that Spirit's e-discovery vendor Legility offers the tool
Used by
Spirit AeroSystems, Inc. (defendant), with its e-discovery vendor Legility and its law firm
Where
Kansas, United States
Stage
In routine use
Date
, when the source was published
Kind
Benefit created
Broad benefit
Processing Information, Getting Information
Collections
Scrutiny and findings
Last revised
1 November 2026

Sources

  1. Court document, by authors with no tie to the system.

    United States District Court for the District of Kansas (Magistrate Judge Angel D. Mitchell). Lawson v. Spirit AeroSystems, Inc., No. 18-1100-EFM-ADM, Memorandum and Order (Doc. 335), 9 April 2020. CourtListener RECAP copy of the filed order.

    In this case, the court finds that Spirit's TAR was reasonable and production of the residual TAR documents is not proportional to the needs of the case.

Details

AI system

Workflow

Spirit ran a TAR review of the 322,000 documents that the plaintiff's search terms and custodians had returned. Predict ranked documents by likely responsiveness and learned continuously from the coding of reviewers. Legility's contract attorneys did a first-level review, and Spirit's law firm did a second-level review.

Follow-up

Conditions for replication

Legility validates the results of the review statistically. The court judges recall against the facts of the case, with 75% to 85% typical in recent guidance. Further review must be proportional to the needs of the case.

Entry details

Sector

Legal services / e-discovery.

Purpose

A developer that the orders do not name built Predict to rank documents by likely responsiveness, learning continuously from the coding of reviewers.

31 more questions not yet checked

model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.

Cite this record

Peter H. Vartanian. (2026). Court accepts Spirit's Predict review stopped at 85% recall (AIBD-0985). Artificial Intelligence Benefits Database, edition of 1 November 2026. https://petervartanian.xyz/aibd/records/AIBD-0985.html (archived at https://doi.org/10.7910/DVN/AADSII, version 7.0)

BibTeX Data Field definitions