What is AI ethics?
The AI Glossary
In our classes, we often pause to explain the ideas behind AI and the language used to describe it. These posts grew out of those conversations. They’re for anyone who would like to understand AI better, with no technical background assumed.
AI ethics examines what people and organizations ought to do when developing and using artificial intelligence. It concerns values such as well-being, autonomy, fairness, privacy and accountability. These are questions about the purposes and consequences of a system, not only whether it functions as specified.
A technical team can measure a model’s errors without settling which errors are acceptable. A service can become more efficient while making it harder for some people to obtain help. An application can follow the law and still raise a reasonable concern about dignity or consent. Ethical analysis helps make those tensions explicit.
There is rarely one number that resolves every value. Giving more people access to information may conflict with keeping information private. Requiring review of every action may reduce some errors but create delays or make a service impractical. The task is to examine the trade-off and its distribution, not pretend that a slogan removes it.
People affected by the system can reveal concerns that designers miss. This may include patients, frontline staff, caregivers and communities not well represented in the available data. Participation should influence decisions rather than merely endorse a project after its main choices are fixed.
Ethics also asks who defines success. A model may optimize an organizational target that differs from the person’s needs. A higher response rate, for example, does not tell me whether repeated messages are helpful, intrusive or confusing.
Consider an AI system that prioritizes patients for outreach. Should it favor people most likely to respond, those with greatest unmet need or those most likely to benefit from the available intervention? These are different aims. The available service capacity and the consequences of being overlooked matter as well.
Ethical reasoning should lead to concrete design choices: what the system may do, what information it needs, how people can challenge a decision and what evidence would justify stopping. The point is not to attach a general statement of good intentions to AI. It is to improve the decisions through which the technology affects people.
Sources: WHO — Ethics and Governance of Artificial Intelligence for Health (2021); Panch, Mattie and Atun — Artificial Intelligence and Algorithmic Bias: Implications for Health Systems (2019)
To learn more, take a look at the AI in Health Care Certificate of Specialization from the Advanced Learning Academy at Harvard T.H. Chan School of Public Health. You can take an individual course or work toward the full certificate.
AI is a fast-moving field. This explanation reflects our class discussions and related material at the time of writing. Terminology, capabilities and tools may change.
For suggested additions or corrections, please email me at tpanch@hsph.harvard.edu.
Dr. Trishan Panch
Program Director, AI in Health Care
Harvard T.H. Chan School of Public Health — Advanced Learning Academy
Lead with AI in Health Care
The AI Pragmatist is a monthly newsletter from ALA faculty, with new explainers & essays from the Advanced Learning Academy