Continuing ed with AI

American Academy continuing ed says at least 6 CE hours per year must be from “organized activities” that involve interaction with actuaries or other professionals working for different organizations.

I just spent some time on an actuarial project with one of the generative AI chatbots. We certainly interacted back and forth. The AI is pretty knowledgeable about state rating laws and actuarial calculations, although it is debatable whether it is a professional. Note that the other professional does not have to be credentialed and does not have to be an actuary.

Was this an organized activity for continuing ed purposes?

My hot take: no. Actuaries and other professionals is a subset of people. AI is not a professional, because it is not a person—it is a tool. Even if it were a professional, it has to be a professional working for a different organization. AI doesn’t work, it’s just a tool that is used by workers.

I’m of course open to more experienced actuaries disagreeing with me. I would recommend any actuary actually wanting to use AI as organized CE (rather than it just being a thought experiment) to reach out to the ABCD for counseling on the subject. At least that way if someone ever questioned it you could point to the ABCD guidance as why you took that course of action (assuming you followed the ABCD’s suggested course of action).

2 Likes

That’s a question that might be worth posing to the Academy.

My $0.02 is no, based on this passage from the USQS:

(emphasis added)

I don’t know that I could claim that an AI chatbot is a “professional working for different organizations” with a straight face.

I wouldn’t count using a search engine to research a relevant topic as an “organized search engine”…and really, with the current state of the art, isn’t engaging with an AI chatbot effectively the same as using a search engine?

I could imagine some day in the future using a prompt of “make a formal presentation to me about current actuarial research on [some subject]” with an AI might produce results as good as attending such a presentation at a seminar/webinar. When we get there, there will definitely need to be some clarification to the USQS.

But I don’t think today is that day.

I’d consider talking with an AI similar to reading CE materials solo as they are tailored for you by a third party.

Unless we hit a point of considering AI capable of making actuarial decisions and taking on legal liability. At that point, my opinion may change.

2 Likes

I don’t think that AI is capable of evaluating its own “knowledge”.

When I was teaching math to high school and college students (one-on-one), an exercise I will use when a student was answering a question, I would ask “are you sure?” even when they gave a correct answer. I wanted the student to also build confidence by self-reflecting on their own work. Once they got use to this tactic, the student would start to automatically correct themselves after giving an incorrect answer.

It may be very good at synthesizing information from a wide variety of sources; but until it can pass the “are you sure” test, I don’t think it can be counted as a “professional”.

1 Like

Good point, to where I’d like to amend my statement…

“similar to reading CE materials solo that were created for you by an intelligent third party who read the materials

Edit: AI does seem to have some ability to evaluate its own knowledge, or at least, it can sense when it’s “guessing” at vague memory versus “knowing” versus “looking it up to verify”, etc. But that feels like a topic I’m unqualified to dive into.

2 Likes

InpYP
:stuck_out_tongue:

I nit-picked Your Post

1 Like

Thanks for all the responses so far.

I agree that AI is not capable of evaluating its own knowledge. We did converse back and forth, and AI did make some corrections. I was working with a simulated dataset, and we performed calculations on it, in real time, and we discussed various intermediate results.

My original question asked if this was an organized activity. I organized it. I may not have publicized it in advance or publicly, but I chose the topic, the time, and the audience. The audience I chose is part of a different organization than mine.

Just to clarify on what it means to be an organized activity, suppose I change the original scenario to having a lengthy private telephone conversation with my company’s consulting actuary. “Mary, I need about an hour of your time around 1 pm for your advice in my complying with the new rating law in the state of X.” I chose the topic, the time, and the audience. This is different than an informal, no particular topic, lunch. Mary is a professional working for another organization. Would this be an organized activity?

By the way, who is the official advice giver on CE questions like this? I don’t think it is ABCD.

On your original question, which audience was present and part of a different organization from you? Did you have the opportunity to interact with that audience? I’d been under the impression you were the only one there.

The AAA website says the ABCD can handle CE questions like this. From Standards & Guidance - Actuary.org “In addition to considering cases involving possible violations of the Code of Professional Conduct, the ABCD provides confidential guidance to actuaries who ask for help in interpreting the Code and/or actuarial standards of practice (ASOPs) and U.S. Qualification Standards.” (emphasis mine)

1 Like

Professionals have standards; AIs do not.

When I talk with myself, it doesn’t count as discussing with other professionals.

And it’s definitely not organized.