In addition to thinking about things and doing things, as Gentlemen Scientists we do also like to talk about things. Having each spent half a lifetime honing our skills in wildly misinformed speculation and meandering discussion, we now feel qualified to unleash some of our thoughts upon the world at large. Each week (or whenever) we will be posting a discussion based loosely around a topic that we have agreed to in advance.
This week we started off with the Turing Test and its history. This then gets us talking about what it means to be intelligent and how much our own human biases might be getting in the way of the search for Artificial Intelligence. Would we know non-human intelligence if it was right under our nose? After all, if and when we ever create/encounter intelligent creatures or machines other than ourselves, they are unlikely to look much like Astro Boy (we love Astro Boy).
We finish up by exploring a novel idea – perhaps instead of trying to teach machines to think, we should be trying to create machines that can feel? Our discussion ends with some speculation about how we might build little machines that use human emotional content as raw material and interact with real people.
*APOLOGIES for the terrible audio. No doubt the quality of audio (and conversation) will improve in weeks to come.
1The Turing Test was proposed by Alan Turing as a way to ascertain whether a machine has attained intelligence. If you are chatting with a computer and you can’t tell whether it is human or not, then the computer has passed the Turing Test – http://en.wikipedia.org/wiki/Turing_test
2TicBot. Artificial Intelligence by accident, as the Guardian puts it.
3ELIZA was an early “A computer program for the study of natural language communication between man and machine”. I wrote my own version of ELIZA from a BASIC listing in an old computer programming book. For some reason I called it KRONOS. Here is the BASIC program listing.
4Many computer programs claim to have “passed” the online Turing Test in competition, but none of really have. Mostly what they have done is approximate human conversation slightly better than their competitors. Then again, maybe we just don’t give them credit because we keep ‘shifting the goalposts’ like in much of AI.
Chatterbots and the online Turing Test – http://en.wikipedia.org/wiki/Chatterbot
The Loebner Prize, Turing Test competition – http://www.loebner.net/Prizef/loebner-prize.html
5Extra credit for Turing Test 🙂 – https://xkcd.com/329/
6The Chinese Room is a famous thought experiment from John Searle. Can a machine speak Chinese by only following mechanical rules/lookups? Because if it can, then we have an intelligent Chinese speaking machine that doesn’t actually “know” Chinese.
7Building a Theory of Mind – http://philosophyofbrains.com/wp-content/uploads/2013/09/minimal-theory-of-mind.nv_.pdf
8Empathetic computing – http://www.cmu.edu/vis/NEW%20WEBSITE/images/publications/empathetic.pdf