Week 9 Studio: Report or narrate? Sorting real ADAS alerts

Students sort real driver-assistance alerts into "reports a state" and "narrates a decision," then argue about which kind is harder to design well

19 April 2027

Before the studio#

Bring one real driver-assistance alert you’ve personally experienced or can find clearly documented — a forward-collision warning, a lane-keep-assist correction nudging the wheel, an adaptive-cruise-control gap alert, an automatic-emergency-braking activation, anything from an owner’s manual, video, or your own car.

In the studio#

Idris hands out a shared list of a dozen real alerts pooled from what students brought in. In small groups, sort every alert on the list into one of two piles: “reports a state” (tells the driver a fact and leaves the decision to them) or “narrates a decision” (describes an action the car is taking, or about to take, on its own). Several alerts are written to sit uncomfortably in the middle — a lane-keep-assist correction, for instance, is a small autonomous action but a barely-perceptible one, and groups usually split on it. Once every group has sorted the full list, compare piles across the room and dig into the disagreements: what made an alert ambiguous, and was the ambiguity in the alert’s design or in the group’s definitions?

Afterwards#

Write up the one alert your group argued over longest and the category you concluded is harder to design well — a report that must be precise, or a narration that must not feel like a malfunction. This becomes your worked example if the mini knowledge check in the week 9 deck calls on it.