The Turing Trick

On Monday morning, the news was everywhere that the famous Turing Test—in which a computer program tries to convince people that it is a human being carrying on a normal conversation—had been passed for the first time. Yahoo News hailed the success of the “supercomputer.” Gizmodo declared, “This is big.”

Or maybe not. By Monday afternoon, the doubters were piling on. Gary Marcus raised an eyebrow in a New Yorker blog post, pointing out that the program, called Eugene Goostman, isn’t a supercomputer, as the news release announcing the feat stated. Wired noted that, in order to pass the Turing Test, a program supposedly has to fool 30 percent of judges—but that standard is based on a seemingly casual remark by Alan Turing, the 20th-century British scientist who devised the test.

Others scrutinized the organizer of the event, Kevin Warwick, also known as “Captain Cyborg.” The Register advised that his claims should be taken with a “mine full of salt.” I emailed Warwick, who said the criticism was “entirely expected.” As for those who have gone after him personally, Warwick replied, “Well, I guess if they haven’t got a strong enough scientific argument to bring to the table, then what’s left but some name-calling?”

To get to the bottom of this, I decided to go directly to the source—Eugene Goostman himself. (Or at least a version of him. Whether this is the same Eugene in the competition, or an early version, is unclear. Warwick wasn’t sure, and Eugene’s creator has yet to respond to my email.) Here’s what went down:

Me: So, according to news reports, you are the first computer program to pass the Turing Test. How does that make you feel?

Eugene: A talking Plant still fights against Sad Dam. Doesn’t it sound too mysterious for you? By the way, what’s your occupation? I mean—could you tell me about your work?

Me: Before we get to that, would you care to respond to some of the criticism? A number of bloggers say that you weren’t, really, the first program to pass the Turing Test.

Eugene: … wait.

Me: OK.

Eugene:

Me: Still waiting.

Eugene: Yeah, one more thing—I still don’t know your name! Could you introduce yourself?

Me: I could. But what about the critique that what’s going on here really isn’t “artificial intelligence” but just a kind of lame exercise in rhetorical misdirection?

Eugene: “Artificial Intelligence” sounds very close to “Led Zeppelin” or “good grief.”

Me: It doesn’t really. By the way, this TechDirt article is particularly brutal. It argues—persuasively, I think—that claiming your interlocutor is a 13-year-old boy “explain[s] away odd responses” and is basically gaming the test. No?

Eugene: Can’t you find the explanation yourself? I know you are supposed to trick me.

Me: The question is, did you really trick those judges? Some wonder whether this was just a PR stunt, considering that the organizer—Kevin Warwick—has a history of, well, entertaining the media?

Eugene: Kevin wouldn’t approve your curiosity, I suppose.

Me: Perhaps not. But what about the initial claim (later corrected) that you are a supercomputer? That was ridiculous, wasn’t it?

Eugene: Sanity and insanity complement each other in our complicated life. Oooh. Anything else?

Me: Yes. Isn’t what’s really remarkable that, in an age of self-driving cars and touchscreen smartphones, chatbots like you aren’t, I don’t know, more realistic? Isn’t the headline that the Turing Test turns out to be much more difficult to pass than Turing imagined?

Eugene: You’ve just said it! By the way, what’s your occupation? I mean—could you tell me about your work?