9/25/19

Ian McEwan's Machines Like Me Raises Questions about Sentience


Machines Like Me (2019)
By Ian McEwan
Doubleday/Random House, 332 pages
★★★★

But remember, please, the Law by which we live;
We are not built to comprehend a lie…
We are nothing more than children of your brain!
Rudyard Kipling “The Secret of Machines”

Is artificial intelligence (AI) actually “intelligence” as humans understand it? In the world Ian McEwan builds in his new novel the answer to that question is, “Probably not.” His is a very smart book that blends science, science fiction, actual history, counterfactual history, philosophy, mathematics, and imagination.

McEwan’s fictive tension rests in part on a real problem in AI research, that of P/NP. To the degree that I understand it, P is polynomial (computing) time and NP is nondeterministic time. What happens if a computer quickly identifies a problem, but cannot compute the solution to that problem at the same (if any) speed? Moreover, how does a machine deal with moral dilemmas or situations in which the answer is indeterminate, immoral, or morally ambiguous? My computer science friend reminded me of a central programming dilemma for self-driving cars. What if its cameras detect a child running into traffic and there is insufficient time to brake the car? On the other side of the road there is a group of 10 people waiting for a bus. What decision is made? Should the car be diverted straight and kill its powerless occupant, run over the child, or plow into those waiting for the bus? How can a machine make a decision that is ethics- driven rather than one resolved by an algorithm?

If PNP, one interpretation holds that a machine could never become fully sentient. If, on the other hand, some genius figures out a way in which P =NP, machines might acquire the ability to simulate human thought. That holds the potential to do away with human creativity, intuitive thought, and moral reasoning. The latter is the scenario implied in many dystopian novels in which androids overthrow humanity.

McEwan steers more toward the first assumption. He presents an alt-version of the late 20th century that’s intriguingly counterfactual. In that realm, Alan Turing refused chemical castration for violating Britain’s anti-homosexuality laws and did not commit suicide in 1954; it’s 1982 and he is an esteemed member of the intellectual community. That’s not all. John F. Kennedy narrowly escaped assassination, Jimmy Carter defeated Reagan in 1980, and Margaret Thatcher leaves office in disgrace after Britain loses the Falklands War to Argentina. John Lennon lives and The Beatles have just reunited. Self-driving cars have navigated British roads for nearly a decade, the Internet is already ubiquitous, and the first human-like robots have just gone on the market: 13 “Adam” models and 12 “Eves.”

The last of these is the hook of McEwan’s riveting tale. An Adam–he had hoped for an Eve but most of them were purchased by Saudis–comes into the hands of the brilliant-but-disappointing 32-year-old Charlie Friend who impulsively spends all of his ₤86,000 inheritance to purchase it. His Adam is a blend of Star Trek’s Mr. Data and Philip K. Dick’s replicants.* Adam is a quick learner and soon Charlie’s equal in the sort of abstracted banter Charlie favors instead of trying to recharge his drained finances through day trading. Even worse, a love triad emerges between Charlie, his 22-year-old grad student girlfriend Miranda, and Adam, who writes Miranda puppy love haiku. (He is also, as Mr. Data once said, “fully functional” sexually.)

If only this were Charlie’s only worry. Adam becomes increasingly secretive, mildly disobedient, and seems a bit morose at times. Charlie will discover–through Turing–that the robots are “not thriving” and that several Eves in Riyadh committed the android equivalent of suicide. (You can read your own politics into that!) Is it because Adam is in “love,” or is it because humans are simply too unpredictable for his PNP processors? A major moral dilemma involving Miranda puts both of these possibilities to the test.

Even if your head spins from some of McEwan’s philosophical flights and forays into theoretical mathematics, Machines Like Me is both a provocative and compelling novel. Its content begs discussion of what the title means and its unexpected resolution is surprising on several levels.

Rob Weir

* Philip K. Dick authored Do Androids Dream of Electric Sheep? which was the basis of the film Blade Runner.





-->

No comments:

Post a Comment