Skip to content
Home / Blog /

AI Do’s and Don’ts for Agents

Image of a person's brain with AI

The use of artificial intelligence (AI) continues to dominate the news headlines—and for good reasons. AI’s potential appears as transformative as the dawning of the internet era was on businesses and the society at large. Yet, while the business applications of AI are being digested, its evolving capabilities seem to be running ahead of society’s ability to apply them.

A subset of AI, generative AI can also create images, text and audio, in addition to processing information. Further, there are other AI research models such as DALL-E3, which provides more refinement and nuanced responses to prompts, and Sora, a text-to-video model that is currently in a test phase, which will provide dazzling videos based on the user’s prompts.

Insurance companies are currently focused on AI’s capabilities for underwriting and claims efficiencies while most agencies have focused on AI’s chatbot capabilities of creating content and providing customer support. There are several do’s and don’ts to keep in mind as your agency’s use of AI evolves.


The Do’s

1) Start your agency’s AI journey. AI may sound intimidating, but it’s not if you think of AI the same way you did when you experimented using PowerPoint and Excel. Start out in small steps, considering how AI’s improved knowledge, faster results or better outcomes can help your agency.

Simply type in a question—a prompt—and gauge the results. Then “rinse and repeat” as you become more comfortable using the tool.

2) Remember to be discerning with your content. More than ever, content must resonate in an authentic manner because AI chatbots like ChatGPT will further open the flood gates for even more stilted content. To ensure your content has an authentic tone, the prompts you use should identify the source, for example “I am a local independent insurance agent,” and the intent, for example “Provide an invitation to someone who recently moved to the area to get together.” Additionally, you must use relevant context, for example “To discuss how flooding has been occurring more often in this neighborhood” with a more detailed example such as “Several houses on your street experienced flooding last June.” Finally, you must specify the format, for example “No more than three lines using bullet points.”

3) Understand the differences. It is important to comprehend the differences between using a search engine like Google, which is a knowledge-based engine that answers questions such as “What is the U.S. population?” versus ChatGPT, which is a reasoning-based engine which offers solutions to questions such as “I am working remotely. How can I best soundproof my home office?.” Keep in mind that Google can provide current data, but through spring of 2024, ChatGPT does not.

4) Tailor your prompts to make ChatGPT sound like your own voice. Make the tone more conversational and less rigid by putting it in a format that fits, such as “Be concise” or providing a word count such as 80 words or less. The more information you provide about the intended audience, for example “They are motorcycle enthusiasts and love dogs”, the more personalized the content will be.

5) Invest the time to review online videos that provide AI insights and updates. A useful source to review is the daily online AI newsletter, The Neuron. Its co-founder Pete Huang has several helpful videos you can access for free from the homepage. As he discusses, there are seven basic actions that ChatGPT performs: writing, summarizing, brainstorming, synthesizing, extracting, reformatting and coding. Be sure to learn about all the uses to maximize its utility.


The Don’ts

As is the case with most technology or tools, there are some things not to do when using AI. Here are five of them:

1) Don’t assume ChatGPT is the best solution for all tasks. If it takes more time to write the prompt than to generate your desired response, then compose the email or text rather than spending more time to use ChatGPT as a crutch.

2) Don’t forget the current legal issues. Litigation surrounding whether AI systems violate the rights of copyright holders is ongoing. Be sure not to plagiarize source material that AI generates. This can be accomplished by properly attributing a quote or statistic to the source using a footnote or hyperlink when appropriate. You can also use a free online plagiarism checker, such as Quetext.

3) Don’t fall prey to the myth that ChatGPT can generate advertising copy. For example, avoid phrases such as “We have the lowest prices,” or “We guarantee a perfect fit” for E&O reasons. For independent agencies, explain that “We have a professional staff to listen to your needs, and we offer a choice of companies” or other value propositions, which is a better way to position the firm.

4) Don’t forget to minimize your professional liability exposure. Users of generative AI should remember that it is not infallible and can often provide incorrect answers. Rather, proofreading and fact-checking are essential to ensuring that false information is not merely regurgitated.

5) Don’t sit on the sidelines too long. Carve out time to learn the AI basics and be sure to involve other agency staff—especially those who have an affinity for learning—to brainstorm on practical uses. Remember AI is a tool, albeit a rapidly evolving one, that can bring greater efficiencies and capabilities that could not be imagined just a few years ago.

Kasey Connors
VP of Marketing Operations, Trusted Choice