18389 Chap.4 Technology as a Moving Regulatory Target
Technology as a Moving Regulatory Target
Regulating before we know
Lecture 2 drops the assumption that the regulator already understands the problem.
Its running case is the arrival of driverless taxis in London, where supervised rides began in September 2026, and it asks what anyone would need to know before getting in: safety, the position of people outside the car, liability and data.
Many seats, many forces
Passengers, disabled travellers, cyclists, hire drivers, insurers, boroughs and investors each see the service differently, and the lecture names the tensions between them, from speed against safety to growth against trust.
Social values, markets, policy goals and local street conditions pull in different directions, which is why this is presented as a wicked rather than a tame problem.
Reframing and the uncertainty map
The chapter shows how to turn a yes or no question into a governable one about users, places, conditions and review, and it maps uncertainty across four layers: the machine, people, the technology's trajectory and society.
Testing mostly reduces the first layer; the rest is learned through deployment, which can itself cause the harm a regulator fears. A worked case applies the method to drone deliveries.
What this chapter covers
- 01
What you would want to know before getting in
- 02
Supervised, testing and intention distinctions
- 03
Authorise, wait or authorise under conditions
- 04
Who is in the room and the four forces
- 05
Tame and wicked problems
- 06
The four-layer uncertainty map
Worked example · free
Reframe a question about AI marking in universities
- 1Identify the hidden questions: who may use it, at which times and places, subject to which limits, and what happens if the judgement proves wrong.
- 1Write the reframed question in the conditional form the lecture recommends.
- 1Give one condition for each hidden question, such as human review of failing grades.
- 1Name the review step that keeps the decision open.
Key terms
- Wicked Problem
- A Wicked Problem is a problem with no linear, straightforward answer that has to be managed repeatedly.
- Spatial Heterogeneity
- Spatial Heterogeneity is the variation in streets, users and conditions that one technology meets in different places.
- Uncertainty Map
- The Uncertainty Map is Lecture 2's division of what a regulator does not know into four layers: the machine, people, trajectory and society.
Technology as a Moving Regulatory Target FAQ
Why is waiting not a neutral regulatory option?
Because delay affects the company, anyone who would have gained from the service, and the path the technology follows. Lecture 2 compares it to a doctor with ambiguous results: sending away a worsening patient is itself a treatment decision. Perfect knowledge may only arrive after deployment.
What is the difference between a tame and a wicked problem?
A tame problem has an answer key that more data can find. A wicked problem has no linear answer and has to be managed repeatedly, because values conflict, everything is connected, behaviour changes, outcomes surprise and you learn late whether a measure worked.
Which layers of uncertainty does testing reduce?
Mostly the machine layer, whether it works as intended. The people, trajectory and society layers, covering how humans behave around it, what it will become and what it does to jobs, streets and trust, are learned largely through deployment and social experience.
Why does the lecture use driverless taxis in London?
Because the case is current and makes the uncertainty concrete. Supervised rides began in London in September 2026, and a company has said it intends to offer fully driverless rides subject to permission. The example lets students practise separating supervised from driverless, testing from public service and intention from permission.
Assessment move
Sketch the uncertainty map from memory and add one question of your own to each layer for a technology you choose. Then write the lecture's reframed question for that technology and list conditions for who, where, when and what if we are wrong. For practice in seeing the stakes, describe the same decision from three seats: a user, a worker affected by it and a regulator.
Return to this chapter after the precaution and Collingridge chapters, since the London case runs through all three. Practise writing the three distinctions from the London case in your own words without the notes.
Working through Technology as a Moving Regulatory Target in 18389? Sia is AskSia’s AI Law tutor — ask any 18389 Technology as a Moving Regulatory Target question and get a clear, step-by-step explanation grounded in how 18389 is taught and assessed. Read this chapter free, then take your hardest questions to Sia.