Claim your CPD points
At the recent Actuaries Institute Insights Sessions, GenAI in Practice: A Member Sharing Session, participants described planning a seven-night stay in Rome, designing a possum-proof garden and tracking down appliance manuals to repair a dishwasher.
The work-related examples were just as inventive. Whether you’re early in your career or an experienced actuary, the use cases summarised below, and the key barriers preventing wider adoption of GenAI, are likely to resonate with you.
Coding was the most frequently mentioned use by participants. Participants use GenAI to write R and Python for ad hoc analyses, convert code from one language to another, cleanse existing code and to support Excel reconciliations. For actuaries with limited coding experience, it shortens the path from an idea to a working product. Web scraping and troubleshooting automation workflows also came up.
Polishing reports and emails, capturing minutes and actions from meetings, and summarising large volumes of documents were widely used and valued. One participant used GenAI to rewrite a group report so that it read in a single voice, as if written by one person.
Participants are using GenAI to build and update dashboards and accompanying slide decks, often drawing on multiple data sources. Examples range from job search dashboards to internally developed assistants that generate interactive HTML visualisations.
In claims, image comparison helps manage costs: photos of a vehicle taken at the start of a policy are compared with photos at the time of a claim, so that prior damage can be excluded. In reinsurance, renewal submissions arrive in many formats, and GenAI pulls out the key information into dashboards that support underwriters and reduce admin time.
GenAI is proving a capable research assistant. Participants use GenAI to navigate foreign regulatory websites and industry publications, produce English summaries of relevant materials, and cross-check findings using other tools when they cannot read the source language. They also use it to interpret translated capital standards from overseas jurisdictions and track statutory and regulatory changes across markets. The common habit is to ask for sources and verify them.
Hallucination was the most cited concern. Participants noted that GenAI does not always follow instructions, which erodes trust, and that drafts are often wordy and need human adjustment. A few found that refining communications took longer than writing the content themselves. On governance, participants want to understand how outputs vary across runs of the same input, how often hallucinations occur and how severe they are, and how much review effort day-to-day use will need.
Actuaries are used to digging into a calculation when needed. Relying on intermediate outputs from a black box takes some getting used to, and may not always be appropriate. Participants also raised the quality and traceability of data, including not only where it came from but also where the output is going.
Many organisations tightly control usage. Several participants only have a closed environment without web access, have been advised not to use it for sensitive data, or face limits on the choice of models and tools. Locally hosted models could keep data in-house and lower token costs, but processing would be slower and frontier models would be out of reach. Some participants also voiced a lack of trust in the companies developing these tools, along with concerns about bias, since responses can mirror the expectations built into a prompt, and about information being used beyond the original intent.
Enterprise use is often tracked because it carries a cost, which raises the question of how to balance token spend with efficiency. Participants also asked how newer team members will build foundational skills if GenAI handles the basics. Some said they now use it more mindfully after learning about the environmental impact of data centres.
Looking ahead, we’re excited by the possibilities GenAI offers to streamline both our professional and personal lives. Understanding its capabilities and limitations is key to overcoming barriers to adoption. The best way forward is to keep learning, build our confidence and continue exploring how GenAI can add value to the way we work and live.
Not sure how to get started with GenAI? All Members of the Actuaries Institute have access to the AI and Data Science Learning Resource on Canvas. This platform, which is free for Members to access, is continually updated and is designed to be a ‘one-stop shop’ for information relating to AI and Data Science fields. The resource contains a section devoted to GenAI resources, including a survey which helps to point you to the most appropriate resources for your current GenAI skill level.
The views expressed in this article are those of the author(s) or working group named below, and do not necessarily reflect the views of the Actuaries Institute. This work is licensed under a Creative Commons Attribution-NonCommercial-No Derivatives CC BY-NC-ND Version 4.0.
From model governance to machine learning, explore actuarial perspectives on data science and artificial intelligence
Subscribe to Actuaries Digital for free and receive the latest actuarial analysis, research, and commentary direct to your inbox