London-based mathematician Fry (The Mathematics of Love: Patterns, Proofs, and the Search for the Ultimate Equation, 2015, etc.) ponders thinking machines, the trust we put in them, and the implications for the future.
Forget about the singularity: The thinking machines are already upon us, and they make extraordinarily complex decisions, from how to battle cancer to whether to send someone to jail. The central question about artificial intelligence and the algorithms that drive it is whether we can trust them to do the right thing, especially if we are ceding decision-making power to mathematical constructs and probabilities. As Fry notes, algorithms alone can push us into some uncomfortable territory—e.g., the sentencing of criminal defendants, a process that, though perhaps driven by an altruistic wish for truly blind justice, puts members of ethnic minorities at a distinct disadvantage: The poorer and less educated a person, in many instances, the more a risk for nonappearance or flight he or she is judged to be. There may be reasons for that failure to show up in court; for one thing, as Fry asks, “do they have access to suitable transport to get there?” Programming the algorithm to account for “societal imbalances” may be one solution, and AI may be able to get around some of the discrimination that would bias a human judge. Still, programmers are people, too. In theory, technology is morally neutral—a drone can be used to take photographs or to kill people—so what really unfolds is what Fry describes at the outset: “Each [algorithm] is inextricably connected to the people who build and use it.” The author writes ably and accessibly of some of the thornier problems, not just in the administration of justice and health care but also in matters like the Bayesian inferences that go into operating driverless cars safely and using algorithms to revise film scripts to “make a movie more profitable at the box office.”
A well-constructed tour of technology and its discontents—timely, too, given the increasing prominence of AI in our daily lives.