AI Basics for Brokers
Part 1 of The Broker's Guide to AI — what AI is, how to talk to it, and why it sometimes makes things up.
This is Part 1 of The Broker's Guide to AI — a plain-language series for insurance professionals who want to understand AI without learning to code. Each part stands on its own, so you can read this one or work through the whole series.
What is AI, really?
Forget robots. The AI everyone is talking about today is software that has read an enormous amount of text and learned the patterns of how people write, explain, and reason. Because of that, it can answer questions, summarize documents, draft emails, and follow instructions written in plain English.
Think of it like a new hire who has read every book in the library but has never set foot in your office. Smart, fast, well-read — but it doesn't know your carriers, your clients, or your processes until you give it that context.
LLMs: the engine under the hood
LLM stands for “large language model.” This is the actual technology behind tools like ChatGPT and Claude. When someone says “the model,” this is what they mean.
The simplest way to think about it: an LLM predicts what words should come next — the same way your phone suggests the next word in a text — except trained on so much material that it can hold a real conversation, explain a policy wording, or compare two quotes.
Prompts: how you talk to AI
A prompt is simply the instruction you give the AI. “Summarize this renewal letter” is a prompt. So is “Compare these two auto quotes and flag any coverage gaps.”
The golden rule: brief the AI like you'd brief a person. Vague request in, vague answer out.
- Weak prompt: “Tell me about flood coverage.”
- Strong prompt: “Explain to a homeowner in Ontario, in plain language, what overland flood coverage includes and what it usually excludes. Keep it under 150 words.”
Hallucinations: when AI makes things up
Sometimes AI gives an answer that sounds completely confident and is completely wrong. It might invent a policy exclusion, a coverage limit, or a regulation that doesn't exist. This is called a hallucination.
It happens because the AI is built to produce a fluent answer, not to say “I don't know.” Picture a keen new hire who would rather guess than admit they haven't read that carrier's manual.