The Ethical Issues of Predictive Coding

More and more courts today are embracing predictive coding or the use of computer-driven algorithms in e-discovery of large volumes of digital assets (such as scanned documents or emails). In just a few short years, the practice has gone from a futuristic tool viewed with no small amount of doubt to a widely-accepted practice that is not only allowed by most courts but is now actually imposed by many judges due to how effective it has become.

New Issues

But any new technique brings with it a responsibility to consider its impact on the ethics and guidelines that attorneys follow. When you are relying on an algorithm to identify and sort thousands or even millions of assets during discovery, the rules and best practices that were clear and well-defined when everything was done on paper and reviewed by human beings become unclear and confusing. What are the new ethical dilemmas every attorney engaged in predictive coding should consider?

Garbage In, Garbage Out

The number one ethical concern for predictive coding is that the process used must be designed well. Highly-experienced and well-trained experts who review the samples taken from the discovery pool are the cornerstones of a well-designed predictive coding process. For the algorithm's "training" to be effective and unimpeachable, these individuals must have the necessary knowledge to be able to identify and categorize documents.

Review and Involvement

No matter the quality of the initial group of reviewers, any ethical predictive coding process must also heavily involve the attorneys. In addition to being involved to judge legal issues and to maintain the discovery process within the normal bounds of legal and ethical requirement, attorneys must also have the expert knowledge needed to identify and categorize assets. In short, any attorney using predictive coding must be able to defend and describe the process used to train the algorithms. Abandoning this process to salesman and vendors of predictive coding systems is akin to abandoning your moral obligations.


The impact of misidentified assets that are not shared because the algorithm incorrectly identified them continues to be the main issue involving predictive coding. An ethical approach should exert real effort and. It should also employ secondary review and possibly random human review of samples to ascertain the accuracy of the computer's work. The reason for this is that human supervision over the process only has a small chance of identifying some of these lapses if done half-heartedly.

There is no doubt that predictive computing is the future. The work now involves learning how to use this new tool both effectively - and ethically.

(This blog article is based on the article entitled "The Ethics of Predictive Coding" written by Robert Hilson published last June 10, 2014 at

James F. Page is a certified electronic discovery specialist. He is also approved as a certified mediator and has been helping individuals throughout Florida settle their disputes using mediation. Call 407-341-0069 to learn how mediation can help you or visit for more information.

This article was published on 26 Jul 2014 and has been viewed 591 times
EasyPublish™ - re-publish this article for free
Featured Slideshare