The M-CaReM-AI Method: Defining Work for AI / Unit 1 MCM 021 review, 2 to 3 minutes
A Library, Diagnosed
The idea
A worked example, constructed for this course, runs the six-question diagnostic on a small public library's loans system.
In practice
The library lends books, films and laptops. Asked for its most important things, the team lists Member, Item, Loan, Hold, Fine and Branch. Question two, whether everyone would agree, gets the first "it depends". The front desk counts a renewal as the same loan with a new due date, but the reporting system counts it as a new loan, so the monthly figures never match. Question three finds no single place that says what a Loan is. Its fields are split between the circulation screens, a policy page and the head of circulation's memory. Question four fails too: a new staff member is told to shadow someone for a week. Question five shows a recent change, longer loans for laptops, that broke the overdue notices, because nobody knew the notice template read the loan period. Question six finds the library's new chatbot answering members' questions about fines from a general description, not from the library's own rules. The team writes four backlog lines. Define the Loan, including whether a renewal is a new loan. Write down the Fine rules in one place. Trace the Loan to everything that reads its loan period. Give the chatbot the library's own definitions.
Where people slip
Every one of the four lines names a thing. The library did not have a staff problem or a software problem, though it had been treating each failure as one of those. It had no agreed account of what a Loan is, written down where everyone could find it, and every failure the diagnostic found followed from that single gap. Fixing the Loan first will make the other three lines easier.
Your place on the trail
Every lesson rests on earlier ones and opens later ones. Tap any to jump.
Something wrong in this lesson? Flag it · Sources for The M-CaReM-AI Method: Defining Work for AI