Capstone: Designing an Ethical AI Policy
If you were in charge, what rules would you set for AI?
Over the past nine lessons, you've looked at AI from a lot of angles: how it learns, how it can go wrong, who it can hurt, and how it's reshaping work. Today you put all of that together and do something genuinely difficult — write the rules.
Think about it like designing a constitution for a small country instead of just complaining about its problems. Anyone can point out that an AI system might be biased, might hallucinate, might mishandle privacy, or might displace workers unfairly — you've spent two weeks doing exactly that. It's a different skill entirely to sit down and decide, in concrete terms, what should be done about it before the system is ever built or shipped.
Real organizations actually do this. Companies, governments, and research labs publish "AI ethics frameworks" — documents that try to answer questions like: How transparent must a system be about how it makes decisions? Who is accountable when something goes wrong? What data is off-limits to collect or use? How do you check for bias before launch, not just after the damage is done? These frameworks differ in detail, but they tend to circle back to four recurring pillars: bias and fairness, transparency, privacy, and accountability.
Those four pillars are your rubric today. A strong AI ethics policy doesn't just say "be fair" in the abstract — it says something a developer could actually act on, like "before deployment, test system outputs across demographic groups and document any disparities." It doesn't just say "be transparent" — it specifies what users are told and when. Vague good intentions are easy; specific, enforceable commitments are the hard and valuable part.
There's no single correct answer here — reasonable policies for a hospital AI, a hiring AI, and a social media recommendation AI will look different, because the stakes and stakeholders are different. What matters is that your policy is specific enough that someone could actually check whether a system follows it, and that you can defend why you drew the lines where you did. That's the real skill this whole course has been building toward: not fearing AI, and not blindly trusting it either, but being able to reason clearly about it.
Try It Yourself
Draft Your AI Code of Ethics
Choose a specific context for your policy (for example: an AI tool used in your school, a hiring AI at a company, or a healthcare AI). Draft a short AI Code of Ethics — at least one concrete, checkable commitment for each of these four areas: (1) Bias & Fairness — how will you check for and address unequal outcomes? (2) Transparency — what will users be told about how the system works or makes decisions? (3) Privacy — what data will and won't be collected, and how is consent handled? (4) Accountability — who is responsible when something goes wrong, and what happens next? Avoid vague statements like "be fair" — write commitments specific enough that someone could check whether they're being followed.
What context did you choose, and what is your AI Code of Ethics for it? Address bias, transparency, privacy, and accountability.
Want to go deeper?
What is AI Ethics?
For Teachers: Full Lesson Plan Detail
- Synthesize concepts from both units into a coherent policy
- Draft an original AI Code of Ethics for a chosen context
- Present and defend policy choices to peers
1. Warm-Up
Quick poll: students vote on the single most important AI risk discussed this unit and explain why.
2. Direct Instruction
Review real-world AI ethics frameworks—transparency, accountability, fairness, privacy—as models.
3. Guided Practice
Small groups draft a 5-point AI Code of Ethics for a chosen context, addressing bias, transparency, privacy, and accountability.
4. Presentation
Groups present their code of ethics; class gives peer feedback using a simple rubric.
Assessment: Final Capstone Rubric: policy completeness, use of unit vocabulary, clarity of reasoning, and presentation quality.