AI Ethics — Explain Like I'm 5
The Unfair Hiring Machine
In 2018, Amazon quietly scrapped an internal AI tool it had built to screen job applicants. The problem? It had learned to penalize resumes that contained the word “women’s” — as in “women’s chess club” or “women’s college.”
The AI had been trained on 10 years of Amazon’s historical hiring decisions. During that time, the tech industry hired mostly men. So the AI learned that “man-like” resumes were more successful, and systematically downgraded women’s applications.
Nobody designed it to be biased. It reflected the bias that already existed in the data.
This is the core challenge of AI ethics: AI systems learn from the world as it is, including all its existing unfairness.
What AI Ethics Is About
AI ethics is the study of how to build AI systems that are fair, accountable, transparent, and respect human rights. It’s not just philosophy — researchers, lawyers, and engineers work on concrete problems:
Fairness: Does the AI make worse decisions for certain groups? A face recognition system that works 99% of the time for white men but 65% of the time for dark-skinned women (real data from MIT research in 2018) is not fair.
Transparency: Can you understand why the AI made a decision? If an AI denies your loan application, you should be able to know why. Many AI systems are “black boxes” — they give answers but can’t explain their reasoning.
Accountability: When an AI system makes a mistake — misidentifying someone as a criminal, recommending the wrong medical treatment — who is responsible? The company that built it? The hospital that used it? The person who deployed it?
Privacy: AI systems can reveal things about you that you never explicitly shared. A model trained on your location history could infer your religion (based on which place of worship you visit) or health status (based on hospital visits).
Why It’s Getting More Urgent
As AI systems make more consequential decisions — about who gets a mortgage, which patients get priority medical care, which job applicants get interviews — getting the ethics right has real stakes.
The EU’s AI Act (2024) now legally requires some AI systems in high-risk applications to be audited for bias and explainability. That’s a sign that AI ethics is moving from academic discussion to legal requirement.
One thing to remember: AI doesn’t create bias — it inherits and amplifies the biases already present in data, which is why auditing, diverse teams, and fairness metrics matter more than good intentions.
See Also
- Ai Safety Why some of the world's smartest people are worried about AI — and what researchers are actually doing about it before it becomes a problem.
- Prompt Injection The security vulnerability where AI assistants can be hijacked by hidden instructions in documents they read — and why it's becoming a serious security problem.
- Reward Modeling How AI learns what 'good' means — the training component that translates human preferences into a mathematical score that AI systems can optimize for.
- Rlhf How ChatGPT learned to be helpful instead of just clever — the feedback loop that turned raw AI into something you'd actually want to talk to.
- Activation Functions Why neural networks need these tiny mathematical functions — and how ReLU's simplicity accidentally made deep learning possible.