Can you use generative AI to understand financial materiality?

I have been testing a few well-known tools (but not ChatGPT because, well, Sam Altman)

Key finding: Microsoft Copilot is considerably better than Gemini or Claude.

🛠️ The task: using a highly curated input data set of the 8 most material ISSB topics per FTSE100 company for FY2025, which company's materiality profile is closest aligned to their business model? The same prompt was used in each tool. The first answers were analysed.

🚮 Gemini didn't understand the query, returned a garbage answer.
🦾 Claude understood the query, performed a plausible analysis, returned a fairly silly answer but with great confidence.
🏆 Microsoft Copilot answered the query sensibly and offered a number of useful follow up queries to contextualize the first answer.

🤔 Conclusion: more research to do but this speaks to the trustworthy AI debate. AI is absolutely useful in materiality assessments, but it needs to be used with care and a lot of subject matter expertise.