The Turning test was proposed 70 years ago to determine if a computer had Artificial Intelligence, based on its ability to fool a human into thinking the computer was human.
But does it still stand up as a test?
This may sound a little harsh, but based on much of what I see on Facebook and Twitter, there might be a few humans out there that would fail the test!
The idea of the test is simple - within a limited period of time (originally a few questions), if the questioner concluded that they were communicating with a person rather than a computer, the device was considered to have passed the test.
We have come a long way with computing power since then, and I amsure we have all come across the Chatbots that are used in many front line cusomers service soltuons on the web. They have a limited range, but if you know what you want, the interaction is much the same as with the customer service person. But if you are a bit unsure (or use a bit of humor as I tend to), the transction gets away from it and soon you see the dots as it trys to connect you to a human!
Does that make it AI - probably yes - it would have passed the Turing test, probably because the user interface is much better now.
Does it make it Skynet? No - it's not sentient, and its not self-aware, but it is capable of fooling people by following a communication pattern that we expect.
So, for me the Turing test is the land speed record of 1950 (around 400pmh) - great for its time and astounding, but as the world has changed, the number is now pedestrian (the record is now 760mph - faster than sound!).
Kommentare