CHAPTER 8, C. The Terminator Argument : The Duty of Competence in Using Artificial Intelligence

• Jurisdiction: United States

C. The Terminator Argument¹: The Duty of Competence in Using Artificial Intelligence

ABI Journal

November 2019

Caleb Chaplain

U.S. Bankruptcy Court (W.D. Va.)

Harrisonburg, Va.

Nisha R. Patel²

Samuel I. White, PC

Richmond, Va.

Key players in the bankruptcy system are inundated with daily sets of data on debtors, creditors, assets and liabilities. Reviewing and analyzing this data can take hours of an attorney's day. Bankruptcy software and related programs with artificial intelligence (AI) components can aid attorneys in processing the deluge of information, but as AI is incorporated into the legal world with increasing frequency, attorneys must consider the ethical ramifications of using such programs.

In August 2019, the American Bar Association adopted a resolution that "urges courts and lawyers to address the emerging ethical and legal issues related to the usage of [AI] in the practice of law including: (1) bias, explainability, and transparency of automated decisions made by AI; (2) ethical and beneficial usage of AI; and (3) controls and oversight of AI and the vendors that provide AI."³ Attorneys must undertake an analysis of some of the ethical implications associated with the use of AI in their practice of law. This article plants the seed for discussion regarding the duty of competence.

What Is AI?

Although the term "AI" often conjures images of machines that ultimately turn on the humans who created them, it actually encompasses much simpler, modern programs, such as email filters, Roombas and Amazon's Alexa. Essentially, AI refers to a machine's ability to exhibit some form of human intelligence.

AI has infiltrated the legal profession, although many attorneys might not be aware of either the extent of their current AI usage or even that they are using it. Pattern recognition and natural language processing in particular have produced beneficial tools for legal research, discovery, contract management, compliance and due diligence.⁴ Bankruptcy lawyers often rely on AI through these methods when using various forms of software in their practices.

Competence

Despite rapid advances in technology, the current ethical rules⁵ and duties of an attorney still govern and fortunately might provide a solid framework to address the ethical use of AI. Model Rule of Professional Conduct 1.1, which governs competence, mandates that "competent representation to a client requires the legal knowledge, skill, thoroughness and preparation reasonably necessary for the representation." Comment 8 to the Rule directs attorneys to maintain competence by "keep[ing] abreast of changes in the law and its practice, including the benefits and risks associated with relevant technology."⁶ It follows that an attorney who uses AI must have some level of skill — although not necessarily expertise — with the technology.

Competence is central to the adoption and utilization of AI because an attorney's competence (or lack thereof) might have implications on his/her other ethical duties.⁷ While it is clearly necessary for an attorney to have at least some understanding of AI to assess the risks and benefits of using a particular program, the question remains as to how much knowledge is sufficient to fulfill one's ethical obligation. Stated differently, what level of technological understanding is "reasonably necessary for the representation"?

In overly simplistic terms, most AI interprets input data to formulate responses through a process called "supervised learning." To begin, the system receives training examples of input and related output data, which helps it learn how to make decisions. For an AI to learn to generate even basic responses, however, it requires a large volume of data.⁸ A basic level of competence might therefore be found in being aware of the data used in the creation of the technology.

A primary issue that an attorney confronts in seeking out the underlying processes is the "black box" challenge, in which "a lawyer submits a query to an AI-powered tool, it goes into a 'black box,' and the AI-based solution provides an answer."⁹ Attorneys may not need to understand every detail of the algorithm that led to the AI's determination, but they should know what information underlies the system and what specifically constitutes the programmed goal.

Transparency aids in this quest. Developers should provide detailed information on how the AI is...