Could Twitter be the venue for a real world Turing Test?
Alan Turing is in the news at the moment, but the thing I've been thinking about is the test named after him. The Turing Test is a way of assessing machine intelligence, with the mark of true machine intelligence being when it is indistinguishable from a human.
Two things about machine intelligence. I recently phoned up my credit card company and had a complete conversation with a very effective voice recognition system that said goodbye at the end of the phone call when all the things I wanted to sort out had been sorted. I actually said goodbye back before I realised what I was doing. The other thing was a game where you press the next word suggested by your phone's autocomplete function until you end up in a repeating loop (which you will). Several people (presumed humans) on an online message board posted their autocomplete paragraphs, all of which made some sort of sense up to a point. It read like people's attempts to communicate in another language, where the words kind of made sense but didn't really.
Spambots have been part of the Internet it feels like forever. Some of the spam comments that get posted on my other website are almost convincing. They steal text and manipulate it in mimicry of actual comments. Bots are also the bane of Twitter, fake followers, odd retweeters of old tweets, little unintelligent ghosts masquerading as real people. Recently, a guy I follow boasted that he set up a new Twitter account as an experiment and attracted 10,000 followers in a weekend - almost all of them bots. He then said that despite the bot problem, Twitter was still an excellent place for a "natter".
That got me thinking, we natter on Twitter to people we know, and to some people we only know through Twitter, and we know the difference between a real person tweeting and a bot. We can spot it and differentiate. But for how long?
The problem with Turing Tests is that if you knew you were being tested that would make you suspicious of the conversation you were having. You would look for clues that it wasn't a human on the other end of the conversation. But what if you weren't primed like that?
Twitter actually could be a real world Turing Test testing-ground. Set up an intelligent machine to interact with people and see if people did actually interact with it, thinking it was human. How long would it be before people worked out it wasn't?
There is a whole lot of information in the Wikipedia article about the Turing Test, including the question of whether people should be told whether they are possibly talking to a computer. But if you don't want people to know, I think Twitter will be the place to run it.