Ethics in Practice: Artificial Intelligence. Case and Analysis–Week of 25 February
Check out the analysis to see how you did in analyzing this week’s case (25 February) and determining which CFA Institute Standard was involved.
Marshall Ng launched his independent investment advisory firm last year. His primary investment tool for managing client accounts is an artificial intelligence (AI)-based model that seeks growth investment opportunities while minimizing excessive risk. Over the past decade, Ng honed his quantitative and computer-modeling skills as the lead technologist for a hedge fund. His AI model makes all investment decisions and submits trading requests without any interactions from Ng. To validate the model’s operations, he regularly reviews his clients’ return performance as well as traditional risk metrics.
Ng is constantly seeking new sources of data for his model. Each new source is back tested against the initial model design and its current sources. Those new sources that strengthen the model’s goals are subsequently incorporated into the live model.
Following the addition of the newest data source addressing market sentiment, the model’s results became extremely volatile and its risk metrics increased significantly. Ng believes the volatility and increased risk metrics will be short lived and will ease as the model learns to apply the new data effectively. He is partially correct — the return volatility stabilizes over the next six months, but the risk metrics remain above desired values.
Ng’s decision to use an AI-based model to select investments is
- inappropriate if he does not understand the basis for the AI model’s investment and trading decisions.
- inappropriate if he does not retain the appropriate documentation to support the AI model’s investment decisions.
- inappropriate if he does not communicate to his clients the continuing updates to the AI model.
- all of the above.
- none of the above.
Technological changes are a consistent part of the investment management industry. From the time that computers began replacing calculators and journal ledgers, the industry has used technology to develop practices and techniques that enhance research and trading efficiency. Although artificial intelligence is just the latest iteration of the ongoing advancement of technology, fundamental ethical norms must be applied to its use to ensure that investors’ interests continue to be protected.
CFA Institute Standard V(A): Diligence and Reasonable Basis requires CFA Institute members to exercise diligence, independence, and thoroughness as well as have a reasonable and adequate basis supported by appropriate research for taking investment action. In the realm of Al-based decision making, all decisions are made within the programmatic platform. Ng reviews the model’s performance and risk metrics, but it is unclear from the facts if his validation of the decisions is grounded in sufficient research.
CFA Institute Standard V(C): Record Retention requires CFA Institute Members to develop and maintain appropriate records to support their investment actions. But in Al-based decision making, the process and information used to arrive at specific decisions are within the programmatic platform. From the information provided, it is unclear what, if any, processes are in place to support appropriate decision-based record retention.
CFA Institute Standard V(B): Communication with Clients and Prospective Clients requires CFA Institute members to describe the basis of the investment process. This information allows clients to make informed decisions about engaging with an investment adviser. With AI, the investment decision-making process continues to “learn” and evolve as data are provided to the programmatic platform.
Ng’s introduction of the new sentiment data transforms the initial model used for back testing into the evolved model used in practice. The question, then, is whether the program’s “learning” process is considered a significant change to the investment process that needs to be disclosed to clients. Individuals researching investment options certainly rely on many sources of information. But a human’s ability to consume data is not as great as that of an AI-based model. The outcome described here of the introduction of a new data source demonstrates the model’s potential sensitivity to new factors. It is unclear from the facts if Ng’s clients have been informed of these changes.
Although the use of AI represents an advancement in investment management, all of these considerations must be addressed in some manner as they relate to ethical practices that protect investors. Choice D is the best answer.
To better understand these and similar concerns, the CFA Institute Standards of Practice Council (SPC) issued a consultation seeking input from CFA Institute members and other industry participants who are using or researching AI techniques. The SPC will consider the responses received in the development of future guidance on the Code of Ethics and Standards of Professional Practice.
The consultation is open until 29 March 2019.
Let us know what you think of Ethics in Practice by taking this short survey.
Have an idea for a case for us to feature? Send it to us at firstname.lastname@example.org.
More About the Ethics in Practice Series
Just as you need to practice to become proficient at playing a musical instrument, public speaking, or playing a sport, practicing assessing and analyzing situations and making ethical decisions develops your ethical decision-making skills. The Ethics in Practice series gives you an opportunity to “exercise” your ethical decision-making skills. Each week, we post a short vignette, drawn from real-world circumstances, regulatory cases, and CFA Institute Professional Conduct investigations, along with possible responses/actions. We then encourage you to assess the case using the CFA Institute Ethical Decision-Making Framework and through the lens of the CFA Institute Code of Ethics and Standards of Professional Conduct. Then join the conversation and let us know which of the choices you believe is the right one and explain why. Later in the week, we will post an analysis of the case and you can see how your response compares.
Image Credit: ©CFA Institute