Practical Professionalism
For our second installment of our Practical Professionalism series, we’re going to look at the following hypothetical scenario:
An actuary is told by upper management to complete an auto rate filing with very little turn-around. Since the actuary is already extremely busy with other work, they decide to utilize ChatGPT. ChatGPT creates the filing memo for the actuary along with several other exhibits. The actuary submits the filing without checking ChatGPT’s work.
Candidates are required to abide by the “Code of Professional Ethics for Candidates,” and in the “Code of Professional Ethics for Candidates,” Rule 6 may apply to this situation, which states:
An actuarial candidate shall not disclose to another party any confidential information unless authorized to do so by the principal or required to do so by law, statute, or regulation. Confidential information includes information of a proprietary nature and information that is legally restricted from circulation.
Depending on what type of information the actuary needed to relay to ChatGPT to make the memo and exhibits, they may have violated Rule 6. This same type of rule is found in Precept 9 of the “CAS Code of Professional Conduct” that applies to members of the CAS.
In addition to violating Rule 6 and Precept 9, I would also argue that the actuary is violating Rule 1 of the “Code of Professional Ethics for Candidates,” which states:
An actuarial candidate shall act honestly, with integrity and competence, to uphold the reputation of the actuarial profession.
Since the actuary did not spend the time to check the output provided by ChatGPT, they did not act with the highest integrity and competency. This is also the same verbiage as Precept 1 in the “CAS Code of Professional Conduct.”.
In terms of any Actuarial Standards of Practice (ASOPs) that may apply, since a rate filing is an actuarial communication, the actuary needed to abide by ASOP 41 if they are a member of the CAS in the U.S. Section 3.4.4 states:
An actuarial communication should identify the party responsible for each material assumption and method.
Since the actuary did not disclose reliance on ChatGPT, they are fully responsible for the content created.
There is no question that automation is useful and that ChatGPT is a powerful tool. If the actuary was careful not to use any proprietary or confidential information when using ChatGPT and had taken the time to check the work output by the tool, then they could’ve saved themselves time while still abiding by the rules set forth.