Just thought I’d make some additional points to the previous post. I recently made a rare venture out of my mountain cave to meet my bumbling genius of a friend, who spent a mere hour writing a computer algorithm that analysed a database of poems, calculated the probability that a certain word would be utilised after an initial word, applied this equation to every word in the database and used it to compose an original poem. Naturally, the poem didn’t make sense at all; grammar and sentence structure were not things his algorithm could take into account, and certain words appeared unreasonably often (such as ‘the’) in the computer-generated poem because understandably, these were words that had a much higher probability of being used in poems that were in the database.
However, he pointed out that this was a fairly basic algorithm. With stronger parameters written in, and with a larger database of poems, almost all of these kinks could be smoothed out. Different codes could be written for different types of verse, from rhyming couplets to iambic pentameters. Now I am no world-renowned poet. Theoretically, if a computer had a database of all the world’s poems, it could very well write a better poem than I can.
And that was my friend’s point: collectively, perhaps computers can’t outshine humans because the best they can currently do is imitation and they need us for something to base off of. But if they can imitate the best of us at everything, individually they can beat a human being at anything.
With regards to the Turing test then, or the suggested test in lieu of it, I s’pose the same problem continues to exist. We can’t tell if a computer is capable of intelligence or creativity, because even if they can’t, they can do a damned good imitation of it.
I also realised I seemed to have dissed the entire range of art in my previous post, which was not what I intended. I should have been clearer. It is obvious to me that art which requires skill and aptitude should have value. Certain forms of art, particularly in modern times, do not have such characteristics. I share the suspicions of many others in being sceptical of their aesthetic worth. There is, however, no denying their economic worth. As long as people like and desire it, it is valuable, and some would say it is art. But that is obviously not a metric for deciding if a computer which produced it is intelligent, the results would be far too subjective and erratic.
Just my thoughts, for what they are worth. Have another Hevisaurus song.
One of the comments on youtube was especially poignant: “In the world of pop music, megastar pop-artists continue to churn out the same shallow crap they have for decades. Meanwhile, in a small corner of the metal hemisphere, a group dressed in rubber dinosaur costumes create a gut-wrenching ballad about the last mammoth. Though largely unnoticed, the effect is profound upon those open enough to listen. Another person’s heart breaks.”
Softly Questioning 