The idea
Engineering decisions affect people who never signed up for them: the pedestrian next to a delivery robot, the person whose face it records. Engineering ethics is about how engineers should behave in their professional role: weighing consequences, duties and fairness, and acting when something is wrong.
Three ways to think about it
| Theory | The idea | Watch out for |
|---|---|---|
| Utilitarianism | The right action produces the most good for the most people | Can it justify harming a few for the many? |
| Deontology (duty) | Some actions are right or wrong in themselves | Rules can clash with good outcomes |
| Practical ethics | Decide within real constraints, asking who benefits, who could be harmed, and whose needs count | It’s judgement, not a formula |
The Statement of Ethical Principles
The Engineering Council and Royal Academy of Engineering set out five principles:
- Honesty and integrity
- Responsibility to society (people’s safety and wellbeing come first)
- Accuracy and rigour
- Leadership and communication
- Responsibility for the future (across the whole lifecycle)
Duty of care
The obligation to use reasonable care and skill to avoid foreseeable harm to anyone affected by your work. In practice:
- work within your competence
- if your advice is overruled, explain the dangers
- report suspected danger or wrongdoing
- e.g. IET Rule 14: take all reasonable care to limit any danger of death, injury or ill health
Professional codes (IET Rules of Conduct, IMechE Code of Conduct) set these standards and hold members to account.
Four tests for a dilemma
| Lens | Question |
|---|---|
| Consequences | Who benefits, who’s harmed, and how badly? |
| Duties and rules | What do the law and professional codes require? |
| Fairness | Who carries the burden, and did they get a say? |
| Character | What would a trustworthy engineer do? |
A good answer names the duty, the evidence and the person you would tell.
Real dilemmas
| Situation | Tension | Lesson from a real case |
|---|---|---|
| A manager wants a delivery robot’s speed raised to hit a target | Client target vs safety of people who never chose to share the path | Uber self-driving test car, Arizona 2018: a pedestrian was killed and the investigation blamed safety culture |
| A robot’s cameras record faces, and the team wants to keep footage for training | Better robot vs privacy. Identifiable images are personal data under UK GDPR | R (Bridges) v South Wales Police, 2020: facial recognition use ruled unlawful |
| A known 1-in-500 failure, with a fixed launch date and “we’ll patch it later” | Loyalty and deadlines vs a known public hazard | Challenger, 1986: engineers’ warnings about the O-rings were overruled. Whistleblowers are protected by the Public Interest Disclosure Act 1998 |
A decision guide (Davis)
- Define the problem · 2. Check the facts (and what’s uncertain) · 3. Identify who’s affected and which principles apply · 4. Develop options · 5. Evaluate them (harm, fairness, can you defend it?) · 6. Choose and explain · 7. Review: how do you stop it happening again?
Check yourself
1. A robot passed 30 tests on a quiet route, but crowded conditions are untested. The manager wants a trial next week. What would you do?
One defensible answer: proceed only with modifications: a quiet route and off-peak times, a supervisor, a low speed limit and an e-stop, while gathering evidence for crowded conditions. Name the principle (responsibility to society, accuracy and rigour) and tell the manager the limits of the evidence.