In GOODBYE, KILLER ROBOTS, author Benjamin Branfman uses evolutionary theory to argue that fears about Artificial Intelligence are overblown. Subtitled “Why Artificial Intelligence Won’t Destroy Humanity,” Branfman presents strategic, logistical, and practical reasons why AI is unlikely to control or eradicate humankind.
The 207-page book (including 27 pages of sources) is divided into eight chapters in which Branfman argues three basic points: humankind’s skills evolved for long-term survival; even human weaknesses are strengths; and AI’s advanced mental abilities are not valuable for long-term survival. Applying natural selection, he argues that this is the only reliable mechanism for survival, and an AI that attempted to evolve through upgrades would inevitably self-destruct: “Every time that the AI updates itself successfully, it will inherit a body and mind that were designed by an older and worse version of itself,” he writes. “Consequently, how many times do you think that process can repeat before the update isn’t successful?”
Branfman devotes several chapters to explaining why AI is not superior to humans, why AI will have no motivation to take over the world, and why the world will always need people who think and act without relying on AI. One point is that, because of AI’s superior mental abilities and immortal perspective, “an AI would most likely understand that aggression is pointless—and therefore the AI wouldn’t have any motivation to behave aggressively.” He also notes that many of our fears about AI are anthropomorphic: “We’re assuming that our violent goals and values are so important that an AI would obviously want to pursue the same things.”
Branfman does not claim certainty—indeed, he emphasizes throughout the book that nobody can know anything for sure. He himself makes several questionable claims, such as an afterlife being more probable than not, and that many scientific theories “conflict internally with themselves.” Regarding history, he is particularly unreliable. “Human civilization has never been as tranquil as a history class might suggest,” he writes, even though history classes are often focused on disruptive events. Branfman also fails to quote the AI pioneers and entrepreneurs, such as Geoffrey Hinton, Yoshua Bengio, and Elon Musk, who have been warning about the existential threats they believe AI poses to humanity.
Although some of his opinions are more likely wrong than right, these missteps still do not affect his main arguments. In the end, GOODBYE, KILLER ROBOTS is a well-written and cogent corrective to the fearmongering that dominates the AI discourse.
Written by Benjamin Branfman, GOODBYE, KILLER ROBOTS: Why Artificial Intelligence Won’t Destroy Humanity is a clearly written polemic against AI doomsday scenarios with a novel perspective: that evolutionary theory, when applied to AI, shows why humans are the superior survival machine.
~ Kevin Baldeosingh for IndieReader