WEEKLY INSIGHTS
An ethics case study: AI in investment management
INVESTMENT professionals have increasingly embraced artificial intelligence (AI) to augment and improve the investment management process in many ways. However, as with any cutting-edge investment practice, managers must be diligent in ensuring new practices do not run afoul of the fundamental ethical principles of integrity, transparency, competency, diligence, and protecting client interests.
The following hypothetical case study and accompanying commentary illustrate a specific use case of AI in the investment management process and related client communications for the benefit of investors.
The case study
Simmons runs an investment management firm serving retail and high-net-worth investors. Eager to take advantage of AI technology and capabilities to enhance his advisory services, Simmons engages the services of a Big Data provider to gain access to raw information that can be used to identify attractive investments.
One area of potential investment that Simmons would like to explore is global food service companies. The Big Data service provider promises a variety of sales data from companies all over the world, including online orders, delivery, app usage, and other information that sheds light on company performance. Simmons plans on mining this data to gain insight into fast-growing regions and products to identify promising investment opportunities in the food service sector.
While the Big Data provider promises comprehensive data for all regions, the data is spotty from food service providers in certain regions. The Big Data provider uses formulas and algorithms to create model data for those regions in which data is incomplete and inserts that model information into the data provided to Simmons.
Simmons hires an AI technology consultant to examine the data to identify global food services companies that could experience above-average growth and profits in the next three to five years. Simmons relies on the consultant to analyse the data and make recommendations. Simmons is excited to use the consultant’s recommendations to enhance client returns.
After the analysis is complete, he invests in the recommended companies for all clients who have given him discretionary authority, regardless of their investment mandate. He then sends the research and recommendations to his remaining clients who retain investment approval.
Simmons is not enthusiastic about sharing his firm’s use of AI to generate research and recommendations because he knows that many of his clients are reluctant to incorporate AI into their investment decision-making process. Simmons is also concerned that disclosure of the role of AI in firm research will cause his clients to question the firm’s robust fee structure because he promotes his firm’s services as “tailored, hands on” investment advice provided by experienced investment managers, all of whom hold the CFA designation.
In conjunction with his efforts to use AI to supplement his investment research, Simmons works to establish a client communication portal for his firm that relies on AI technology. Because he has limited firm investment personnel, Simmons leverages AI technology through an online chat function to keep overhead costs low and improve communication with clients.
The chatbot function adopted by Simmons’ firm is used to distribute new investment recommendations to clients. It is the first line of communication for clients who are seeking basic information about their accounts or who have questions about the firm’s investment recommendations.
Simmons’ goal is to provide more timely, frequent and focused client-specific interactions through AI technology. For client inquiries about research and recommendations on global food service industry investment opportunities, the AI responses reflect Simmons’ enthusiasm for these investments. He programs the chatbot to guide conversations towards promoting the securities of the companies in the AI technology consultant’s analysis, invariably recommending these investments to clients.
Commentary
CFA Institute is a global not-for-profit organisation for investment professionals. The institute has prescribed Standards of Professional Conduct which investment professionals should adhere to.
The CFA Institute Diligence and Reasonable Basis Standard requires one to exercise diligence and thoroughness in analysing investments and acting. While gathering, analysing, and interpreting Big Data may enable a thorough analysis of investments, in this case, the accuracy and completeness of the information in the database is in question.
While the data is supposedly comprehensive data for all regions, the Big Data provider used formulas and algorithms to create model data for those regions in which actual data is incomplete. Use of model data calls into question the accuracy and potential bias of the information. It’s not clear from the facts that Simmons is aware of or takes account of this limitation when using Big Data to analyse the investments and make recommendations to clients.
If the Big Data company’s disclosures relating to the database provided to Simmons do not include the fact that model data is used to fill gaps in actual observations, that would violate the Misrepresentation Standard. This standard prohibits one from making any misrepresentations, including misleading omissions, relating to investment analysis or actions.
The Competence Standard requires one to act with and maintain competence necessary to fulfil their responsibilities. It is clear from the facts that Simmons relies on an AI technology consultant to examine and analyse the data to identify favourable investments and then passes on the consultant’s recommendations to clients. It is not clear that Simmons understands how the consultant uses AI to conduct the analysis or make the investment recommendations.
The Competence Standard does not require one to have expertise in all aspects of the investment process. Hiring an internal team or third-party consultant to fill a knowledge and skills gap is a legitimate way to competently provide professional services to investors. But as the investment manager, Simmons must have at least a thorough, basic understanding of the methods and process for making investment recommendations. Simply passing on the recommendations of a consultant to clients without understanding how the recommendations are formed would likely not meet the competence required by the standard.
Once received, Simmons’ use of the AI-assisted recommendations was also questionable. The facts state that Simmons invested in these recommended companies for all clients who have given him discretionary authority, regardless of their investment mandate. Similarly, he passed on the recommendations to all non-discretionary clients.
There is no indication that Simmons made a suitability analysis to determine whether the investment recommendations were suitable to each client’s financial situation, objectives, mandates, and constraints as required by the Suitability Standard. Simmons’ distribution of the AI-assisted investment recommendations was also flawed.
The CFA Institute Fair Dealing Standard requires one to deal fairly and objectively with all clients when making investment recommendations or taking investment actions. By acting for his discretionary clients before distributing the recommendations to his non-discretionary clients, Simmons treated the latter unfairly.
Simmons’ communication with clients regarding his firm’s use of AI in the investment management process is incomplete and misleading. Simmons is concerned that his clients will not be comfortable with the use of AI to manage their investments or will question his competence or expertise if he shares that he relies on AI to help him in the investment-decision making process. He is also concerned that the use of AI will seem to contradict his claims that he takes an active role in managing client assets and will lead to questions about fees. As a result, he did not disclose the use of AI in the investment decision-making process to his clients.
The Communications with Clients Standard requires one to disclose to clients the basic format and general principles of the investment processes they use to analyse investments, select securities, and construct portfolios. Omitting information on the nature and extent that Simmons used AI in the investment process would likely be a violation of the Communications with Clients Standard, and possibly also of the Misrepresentation Standard.
The same transparency questions arise with Simmons’ use of the chatbot for communication with clients. It is not clear from the facts that clients are aware that communication from the firm when they inquire about their account or receive new investment recommendations is AI-generated and not from the firm’s investment personnel. Transparency dictates that clients be made aware of this use of AI technology in handling their accounts.
Another potential ethical issue with the use of the AI client communication portal is the bias towards investing in AI-recommended securities when responding to client inquiries. Because of Simmons’ enthusiasm for these investments, he programmed the chatbot to guide conversations to promote the securities, invariably recommending these investments to clients. This was done regardless of the suitability or appropriateness of the investments for clients.
The Loyalty, Prudence, and Care Standard mandates that one exercises reasonable care and prudent judgement to act for the benefit of clients. By interjecting bias and not incorporating safeguards into the chatbot program, Simmons may be in violation of this standard and other standards as well. If given free rein, the chatbot could violate several standards. Use of AI in communication and other aspects of the investment process requires prudence, care, and oversight to ensure compliance with the principles of the CFA Institute Code and Standards and ethical investment practices.
The writer is director of professional standards at CFA Institute
Source: The Business Times