Lesson 3 of 8 · 9 min
Choosing the first case
A first case that is useful right away, in the area where AI really helps.
The first use case sets the tone. If it is useful, the team asks for more. If it produces a visible error, distrust sets in for a long time. Choosing it with the team is therefore the most important decision of the launch.
An experiment run with a large consulting firm showed that AI clearly improved consultants’ work on some tasks, and made it worse on others that looked similar, because they trusted it where it was wrong. The authors speak of a jagged frontier. For a first case, stay well on the right side of it.
- A frequent task, that several people do every week.
- A recognised chore, that nobody will miss.
- A result you can check in a few minutes.
- A harmless error, that can be caught before it leaves the team.
The minutes of the team meeting, sorting incoming requests or the first draft of a standard reply are good candidates. A reply sent to a client without review, or a decision that affects a person, are not.
References
- Dell’Acqua, McFowland, Mollick et al. (2023). Navigating the Jagged Technological Frontier: Field Experimental Evidence of the Effects of AI on Knowledge Worker Productivity and Quality. Harvard Business School Working Paper 24-013. doi.org/10.2139/ssrn.4573321