Over the last several years legal technology has been rapidly advancing, reaching a point where the use of artificial intelligence (“AI”) in legal practice has become a reality.[1] AI assists lawyers with many tasks that have been considered tedious and time-consuming without technological assistance, and allows for work to be done much more efficiently.[2] These tasks which AI helps with include legal research, case outcome prediction,[3] and, perhaps most notably, e-discovery document review – through the process of technology assisted review (“TAR”).[4]
One of the major steps in e-discovery involves “the review of [electronically stored information (“ESI”)]for relevance and privilege.”[5] This step has historically required human review, meaning that individuals had to look at each piece of ESI to determine whether it was relevant to the case at hand and whether the information it contained was privileged.[6] Given the vast amount of ESI in a given case, human review can become very time consuming and costly.[7]
For many firms, TAR has become the solution to the issue of time and cost associated with human review.[8] It is a process through which a computer system “electronically classif[ies]documents based on input from expert reviewers.”[9] What this means in the e-discovery context is that after a user inputs a sample batch of documents with classifications, the software creates an algorithm to sort through all of the documents that are set for review and ranks them by their level of relevance to the case.[10] Since the human user must only review and input a sample batch of documents as opposed to all of the documents, the amount of human time put into the reviewing process is significantly less, lowering the overall cost of the review.[11]
In addition to lowering the costs of review, studies have shown that TAR “yield[s]more accurate results than exhaustive manual review, with much lower effort.”[12] Not only is the effort expended through the use of TAR less than that of human review,[13] the accuracy of TAR is also higher than that of human review, given that, unlike humans, TAR is not susceptible to fatigue or simple human error.[14]
As the use of TAR has become more widespread, inevitably litigation has arisen over its reliability in determining which documents to present in discovery.[15] However, the courts have generally approved of the use of TAR.[16] Perhaps most notable is Judge Peck’s opinion, which states that “[c]omputer-assisted review appears to be better than the available alternatives, and thus should be used in appropriate cases.”[17] Other court opinions, while accepting the premise that TAR is generally appropriate for use in e-discovery, address several questions regarding the details of its use.[18] These opinions discuss whether parties may compel the use of TAR, whether parties may switch to TAR after having begun discovery, whether “cull[ing]the document population before applying TAR” through the use of search terms is appropriate, and the extent to which parties must disclose the details of the sample batch that they use to train the TAR system.[19]
TAR is a tool which, when used properly, can enable a firm to significantly lower costs and improve accuracy in the e-discovery review process.[20] Although there are legal questions regarding the details of TAR’s use in e-discovery, the courts have accepted TAR as a proper method of distinguishing relevant ESI.[21] Along with other AI systems geared toward legal practice, TAR can help streamline law firms, stimulating their growth and promoting their productivity to a new level.[22]
[1] See generally Jason Koebler, Rise of the Robolawyers, The Atlantic (Apr. 2017), https://www.theatlantic.com/magazine/archive/2017/04/rise-of-the-robolawyers/517794/.
[2] Robert Ambrogi, Fear Not, Lawyers, AI is Not Your Enemy, Above The Law (Oct. 30, 2017, 3:00 PM), https://abovethelaw.com/2017/10/fear-not-lawyers-ai-is-not-your-enemy/?rf=1; see also AI vs. Lawyers, LawGeex, https://www.lawgeex.com/AIvsLawyer/utm_source=google&utm_medium=cpc&utm_campaign =US_sch_ai_vs_lawyers&utm_adgroup=56678748710&device=c&placement=&utm_term=ai% 20lawyer&gclid=CjwKCAjwypjVBRANEiwAJAxlIuXO0EJtJahuK7 adW6INvcyv0s_aBH_nQjakMGuYzVIrK kfVpfJ0yxoCf6QQAvD_BwE (last visited Mar. 14, 2018).
[3] See Koebler, supra note 1.
[4] See David Lat, How Artificial Intelligence Will Revolutionize eDiscovery, Above The Law (Jan. 25, 2017, 1:03 PM), https://abovethelaw.com/2017/01/how-artificial-intelligence-will-revolutionize-ediscovery/?rf=1.
[5] EDRM Model, EDRM, https://www.edrm.net/frameworks-and-standards/edrm-model/ (last visited Mar. 14, 2018).
[6] See Thomas C. Gricks III & Robert J. Ambrogi, A Brief History of Technology Assisted Review, L. Tech. Today (Nov. 17, 2015), http://www.lawtechnologytoday.org/2015/11/history-technology-assisted-review/.
[7] See Adrian K. Felix, E-Discovery: Shifting the Costs of Compliance, A.B.A., https://www.americanbar.org/groups/young_lawyers/publications/the_101_201_ practice_series/e_discovery_shifting_the_costs_of_compliance.html (last visited Mar. 14, 2018).
[8] See Thomas C. Gricks III & Robert J. Ambrogi, supra note 6.
[9] Technology Assisted Review, EDRM, https://www.edrm.net/frameworks-and-standards/technology-assisted-review/ (last visited Mar. 14, 2018).
[10] See id.
[11] See id.; Thomas C. Gricks III & Robert J. Ambrogi, supra note 6.
[12] See Thomas C. Gricks III & Robert J. Ambrogi, supra note 6.
[13] See id.
[14] See Philip Favro, Lessons Learned From New Technology-Assisted Review Case Law, Driven, http://www.driven-inc.com/lessons-learned-from-new-technology-assisted-review-case-law/ (last visited Mar. 15, 2018).
[15] See Bob Ambrogi, TAR in the Courts: A Compendium of Case Law About Technology Assisted Review, Catalyst (Nov. 14, 2014), https://catalystsecure.com/blog/2014/11/tar-in-the-courts-a-compendium-of-case-law-about-technology-assisted-review/.
[16] See id.
[17] Da Silva Moore v. Publicis Groupe, 287 F.R.D. 182, 191 (N.Y.S.D. 2012).
[18] See Bob Ambrogi, supra note 15.
[19] Id.
[20] See Thomas C. Gricks III & Robert J. Ambrogi, supra note 6.
[21] See Da Silva Moore, 287 F.R.D. at 191; Bob Ambrogi, supra note 15.
[22] See generally Robert Ambrogi, supra note 2.
