A new novel about AI and the Turing test

by Rob Tiller

Sally’s re-reading Anna Karenina, which seems to me both admirable and exhausting. The recent movie version with Keira Knightley was highly stylized, but reminded me of what I enjoyed about the book when I read it in my twenties. It is rich book, full of feeling and thinking. But it’s long!

As a teenager and young adult, I read a lot of long novels, including ones by Tolstoy, Dickens, Dostoyevsky, Trollope, Elliot, James, and Proust. My “big novel” period was a time when I was coming of age and constructing a particular consciousness. Those big books were part of the process.

Nowadays most of my waking hours are spent working, and there is stiff competition for precious non-work time. I’m still interested, though, in novels, and especially ones that take on issues that haven’t been thoroughly mined out. I just finished one such: A Working Theory of Love, by Scott Hutchins (Kindle edition). It’s about a guy who’s working on an artificial intelligence program designed to pass the Turing test, which is a real competition suggested by Alan Turing.

The Turing test is designed to probe whether machines can think. The challenge is to build a computer that can persuade 30 percent of humans that it is human. (I wrote about a very interesting non-fiction account of the test and artificial intelligence, The Most Human Human, by Brian Christian, here.)

Hutchins’s narrator bases his program on his father’s diaries. After getting the computer to converse coherently, he works on humanizing it by adding emotion and sex drive. As the program improves he has the feeling that his father is coming back to life. This creates an interesting moral dilemma. His father had committed suicide, but the project seems to be denying him his freedom to choose death.

I found Hutchins’s premise thought-provoking, but I ultimately didn’t care very much for his narrator. But he’s where the action is. It’s exciting and terrifying to see how fast robotics and artificial intelligence are transforming the world. The AP did a good overview piece last week, which I recommend highly. As they note (and as I’ve noted before), jobs involving any sort of routine (most manufacturing, transportation, retail, and office work) will soon be gone forever, taken over by robots and AI. This means increasing efficiency and wealth for some, and unemployment and anomie for a great many others.

We’re going to need to re-think and re-size our social programs for a world where humans are not needed to produce most goods and services. This is a daunting task, even leaving aside the extreme polarization of our politics. The shift away from human labor as a process that is the source of economic value and meaning is hard for us to grasp and accept. But we somehow need to provide a safety net for the millions who will be affected.

I’m not prepared to propose a program, but I do have the name for one: the Big Deal. It will need to be bigger than FDR’s New Deal. It will surely involve some sort of cash payments and medical care. I’d also add a work program that channeled redundant workers to activities that would provide them with a sense of meaning and purpose, like caring for other humans.