So, I’ve been seeing a bunch of stories basically saying “Machines can think and feel and love now. They passed a test called ‘the Turing test’ so now they’re real boys.”
I have quite a few problems with this.
First of all, the Turing test isn’t a thing. I mean, yes, it has an entry on urbandictionary.com but, it’s not in the same realm of scientific rigor as “speed of light in a vaccuum has a constant speed;” it’s much closer to a theory saying “some people like ham, some don’t”.
So where’d it come from? Once upon a time, this awesome man named Alan Turing said (paraphrased) “Asking if machines can think is a silly question. How would you even measure that? Now, if you asked if a machine can fool a human into thinking they’re talking with another human, that would be a question that you could measure at least…”
Then a bunch of grad students went “So, if you can get a human to think they’re talking to another human, we will have created artificial intelligence, and in doing so, become gods ourselves?”
Turing said “No. Um, wait, what?? Were you even listening? That’s not even close to…” But it was too late. People made this a thing. They even hold contests and stuff now. They also tell journalists everywhere that this is somehow a marker of our progress towards Artificial Intelligence.
So, is it? Surely you’d have to write something pretty complex to be able to talk to a human, right? Not really. Most of the programs I’ve seen or read are really straightforward. You write some text to it, and then it writes some pre-scripted text back. Some of them are mildly sophisticated in that they’ll remember your name, and if you say you came from Denver, they’ll ask you what the weather is like in Denver etc, but your phone can do that too, only more accurately.
A vast majority of these programs are still canned phrase-responses. So, it’ll scan your text for the word ‘cat’, and if it it finds it, it’ll say “Speaking of cats, my cat is…” and then answer very simple questions regarding the program’s nonexistant cat. If you go off-topic, it will respond with something that would make it sound annoyed, so if you say something like “dog” or “potato” or “libertarian” it’ll give the exact same response: “I wasn’t finished. Can we go back to talking about my cat?”
So, really the accomplishment of these tests is to show how banal our conversations really are, and how little we generally listen to, or engage our conversation partners.
This shouldn’t come as a shock to anyone, and at the very least isn’t newsworthy.