Can a Machine Act Ethically?

Making the right decision takes more than rational thought.

I was among the many people who watched “Watson,” the IBM super-computer, easily beat two former champion players on Jeopardy!. This generated much discussion about whether, in the future, super-intelligent computers would displace humans as decision-makers.

Recently, I’ve also spent time speaking to journalists about the ethics of the case of Baby Joseph Maraachli.

Baby Joseph has a fatal, degenerative neurological disease and was on a respirator in a London, Ont., hospital. The hospital and doctors wanted to remove the respirator, as a result of which Baby Joseph would die. His parents refused to consent to this and wanted him given a tracheotomy, so they could take him home and care for him until he died. The conflict was brought before both the Ontario Consent and Capacity Board and a court, and both of them affirmed the doctors’ request for withdrawal of life-support treatment.

Subsequently, with the help of a pro-life group in the United States, Priests for Life, Baby Joseph was taken to a St. Louis, Mo., hospital where a tracheotomy was performed. At the time of writing, Baby Joseph is out of the hospital and back home with his family in Windsor, Ont. He is reportedly breathing through his “trach” with no ventilator support.

Juxtaposing these two events might seem bizarre, but it occurred to me to wonder what answers “Watson” would give, if we asked “him” questions about the ethics that should govern Baby Joseph’s case. (It’s interesting we call “Watson” him, not it.) “Watson” is far superior to humans when there is a clear, rule-based answer, but most ethical dilemmas do not have such an answer. They are dilemmas for precisely that reason. If it’s clear what is ethical and what is not – that is, it’s a black-and-white situation ethically – we must simply do what is right.

But finding the ethical path when there is an ethical dilemma depends heavily on taking into account the context in which the dilemma exists, and subtle nuances in that context can radically change what is an appropriate ethical response. It’s what we call ethical decision-making in the “grey area,” where we must make value judgments and be able to justify, ethically, the decisions we make.

Such decision-making often involves choosing between competing harms – making decisions in what’s sometimes referred to as “a world of competing sorrows.” And it almost always means that we will have to breach some values in order to uphold the ones given priority. Justifying breaching the values that are not honoured is at the heart of “doing ethics.”

So, for instance, in Baby Joseph’s case, we would start from presumptions in favour of respect for life and for parents’ right to make decisions regarding medical treatment for their child. But withdrawing life-support treatment could be justified where the suffering it inflicts on Baby Joseph clearly outweighs any benefits to him. And parents can be displaced as decision-makers if their decisions constitute child abuse or neglect – which was not true of Baby Joseph’s parents’ decisions. Whether these exceptions apply involves value judgments that must be able to be justified ethically. Here’s where “Watson” hits the absolute limit of “his” capacities.