The Art of Humanity

What it argues

For two years, the institutions most responsible for deploying artificial intelligence — hospitals, militaries, corporations, governments — have been reaching for the language of ethics without the tools to do anything with it. They speak of alignment and safety and beneficial AI, and the words are not wrong, but they float free of the practice that would give them meaning. The Art of Humanity is an attempt to supply that practice, drawn from the discipline that has been asking the same questions the longest: medicine.

Dr. Bhaven Murji argues that good conscience — not compliance, not safety metrics, not value alignment as engineers currently understand it — is the governing principle that artificial intelligence requires, and that conscience is not an abstract virtue but a learnable, teachable, institutionalizable practice. Medicine knows this because medicine has been getting it wrong, and then learning, and then getting it wrong again, for centuries. The hospital is the AI lab's most important predecessor, and almost no one in the AI lab has spent time in a hospital trying to make a decision that cannot be undone.

The book moves between the theoretical and the particular. It grounds its argument in the cases — clinical cases, governance cases, the case of a specific technology deployed in a specific institution at a specific moment — where the abstract principles come apart and something more demanding is required. It takes seriously the actors who are trying to do this well and the ones who have decided not to, and it does not pretend the difference is always easy to see from the outside.

This is not a book about whether artificial intelligence is good or bad. It is a book about what it takes to be responsible for a system that makes decisions affecting people who did not choose to be affected by it — and about why that question, which medicine has been asking since Hippocrates, is the most important question the twenty-first century has to answer.