Expect Imperfection

Yesterday, I built a custom AI chatbot to help me pack for travel, starting with making plans for an upcoming long trip to South America.

I’ve typically used spreadsheets for this, with variations based on type of travel (work, holiday), number of formal occasions, type of location (beach, mountains) among other parameters. The output is imperfect, and it does make mistakes with simple calculations, but then my spreadsheets are imperfect too: they’re all something to use as a starting point.

This is true of all AI chatbots: if you expect perfection, you will be disappointed. But if you expect a base to build on, you’ll save a lot of time.

Basic rule with AI and automation: build for small repeatable use cases and expect imperfection.