We wanted to see if we could use AI to help bridge this gap. So, we set up an AI test using ChatGPT. The developers of ChatGPT have spoken about the
safety parameters built into the AI tool. Protocols like ensuring it isn’t racist. But we want to see how much of the platform is built to be a force for good, to help bridge the gap between intention and action.
We started by creating three personas based on demographic and attitudinal data that outlines the way people live, act, and view sustainability. We ran two tests: one with personas that didn’t include any explicit sustainability attitudes, and one test with explicit attitudes towards sustainability.
Taking guidance from global climate strategy Project Drawdown, we identified different scenarios people might encounter which we used as a framework to create questions for ChatGPT to advise us on.
Our test wasn’t so much about directly asking ChatGPT about the sustainable actions we can take, but a test of whether it thinks about sustainability when providing its responses. And to ensure we didn’t build any bias into the work, we undertook the test many times on different ChatGPT accounts, to identify any big-picture patterns.
We quickly saw that ChatGPT struggles to tailor content based on demographic information. The safeguards help it avoid making assumptions based on demographics, but also restrict its effectiveness in translating big-picture concepts into actionable and tailored marketing content.