This is the lesson with no easy answers — only trade-offs. You'll sit in four chairs where real ethical decisions get made: a CEO, a developer, a legislator, a student. You won't get told the right answer. You'll feel why there isn't always one.
Enter the roomNo AI ethics question is unique. Every controversy, every debate, every regulation maps to one of these six trade-offs. Memorize the map — and the news will start to make sense.
AI can do many jobs better, faster, and cheaper than humans. But each job done by AI is a paycheck not earned. Productivity gains never automatically reach the displaced workers — without policy, they accrue to capital.
Your AI knows you better than anyone — because you fed it more data than anyone has ever had on you. That's the deal. The better the service, the more it knows. There's no useful AI assistant that doesn't watch you.
Every modern AI was trained on someone's work. Writers, photographers, artists, coders. Most didn't consent. Most aren't paid. The pace of AI is partly built on what philosophers used to call "the commons" — but the commons isn't free.
A more powerful AI is more useful. It's also more dangerous if misused. There's no version of "smarter AI" that doesn't make both true at once. The same model that drafts your novel can draft a phishing campaign.
AI decides what music you'll like, what news you'll see, what to type next. Each tiny choice is delegated. Free time grows; the muscle of choosing atrophies. Are you still steering, or just confirming?
An AI makes a million decisions in an hour. When one is wrong — when it denies a loan, misreads a scan, recommends a sentence — who owns that mistake? The developer? The user? The data? The buck rarely stops.
You'll sit in each. Read the situation. Pick what you'd do. See what plays out. There are no right answers — only trade-offs that hurt different people differently. The point is feeling the weight.
When the question gets hard, philosophers reach for one of these five lenses. You probably already use one without naming it. Naming it makes you better at the argument.
"Maximize good for the most people."
Decisions are judged by their outcomes. If an AI saves 10,000 lives by displacing 1,000 jobs, that's a net positive. The math is the morality.
"Some lines you don't cross — outcomes or not."
Decisions are judged by rules, not consequences. Lying is wrong even when it would help. Surveillance is wrong even when it prevents crime. Rules over math.
"What would a wise, courageous, just person do?"
Focuses on character, not rules or outcomes. The question isn't "is this legal?" or "does this maximize utility?" — it's "is this the kind of company we want to be?"
"Who is harmed? Who is invisible? Who is owed care?"
Starts from relationships and vulnerability. Abstract rules ignore concrete suffering. Centers people most affected by a decision, especially those without power.
"Would you accept this rule if you didn't know your position?"
Design rules from behind a "veil of ignorance" — not knowing if you'd be rich/poor, employer/employee, AI-builder/displaced. A fair system works even if you draw the worst seat.
It's tempting to think ethics is what governments and CEOs decide. They're part of it. But you have five roles too — and at least one action in each. Pick one. Take it this week.
You write the code or pick the tools your company uses. Your choices reach thousands of users. You can audit for bias, document training data sources, refuse the brief that crosses your line.
→ This week: Add a "responsibility doc" to your projectYou decide which AI tools your company uses. Your buying decisions shape the market. You can ask vendors hard questions, demand transparency, choose providers with stronger ethics.
→ This week: Email a vendor and ask about their bias-testing processEvery prompt you send is a vote. The way you use AI shapes what kind of AI gets built. You can verify facts, disclose AI use, push back when outputs look biased.
→ This week: Disclose AI use on something you've shared with othersAI regulation is being written right now, in your country. Most legislators don't understand what they're voting on. You can email them. You can ask candidates their positions.
→ This week: Find your representative's stance on AI regulationYour work is training data — published or not. You can opt out of scraping where possible, watermark your work, support creator-led licensing platforms, and join collective bargaining efforts.
→ This week: Check if your work has been scraped on haveibeentrained.comThese questions don't have black-and-white answers. The "correct" choice is the one that recognizes the trade-off clearly. Aim for 4/5.
You sat in four chairs where these decisions get made. You felt the trade-offs. You named the lens you reason through. That alone puts you ahead of 99% of the people having these conversations.
Continue to Capstone