ENGINEERS’ DREAMS By George Dyson – Speculation About Artificial Intelligence, & Google’s Part In It
There’s a provocative very short story – ENGINEERS’ DREAMS by George Dyson – that was turned down by the technical press for being too fictional, and by Wired magazine for being too technical.
“Only one third of a search engine is devoted to fulfilling search requests. The other two thirds are divided between crawling (sending a host of single-minded digital organisms out to gather information) and indexing (building data structures from the results). Ed’s job was to balance the resulting loads.When Ed examined the traffic, he realized that Google was doing more than mapping the digital universe. Google doesn’t merely link or point to data. It moves data around. Data that are associated frequently by search requests are locally replicated—establishing physical proximity, in the real universe, that is manifested computationally as proximity in time. Google was more than a map. Google was becoming something else. … “
So it found at home at Edge magazine – click through to read the story before proceeding further as there’s a bit of a spoiler below…….
Mr Dyson’s thesis in this, is that if the huge pile of home & office computers, servers and fibre optic cable that comprise the internet is ever to become alive and attain some kind of artificial intelligence (as has been written about before in many science fiction stories and movies) then it’s going to start becoming intelligent by first of all dreaming, not thinking.
And this is where the problem lies. When I dream, I mostly dream in color, though it’s sometimes greyed out, but never totally black and white. But it’s always pictures. Sometimes it’s conversations and words, though mostly they don’t make much sense when I look at them in the cold light of morning when I awake thinking “What the hell was that all about?”. And just once have I dreamt about numbers, when after trying to solve a particularly knotty coding problem that had been confusing me for days on end, I dreamt in computer code one night (I dreamt in Amos Basic on a Commodore Amiga if you must know).
We first encounter our world with the totally visceral senses of touch, smell, hearing, taste and sight. The sense of language, and the sense of text, is something that only comes along many years later.
But the internet, and Google specifically, get almost all their information from text and to a lesser extent numerical informnation. Google does not understand pictures yet. Google is not going to dream in pictures until some clever person comes up with an algorithm that can break up a picture into its individual components and then give meaning to those components. And this is proving to be a very hard task to do.
I rather suspect that Google has been trying a new algorithm of their own over the last few months. There was a brief trial (can’t remember the name of it) of a picture comparison algorithm where two people would be shown the same picture and asked to describe it using words or short phrases. If you both came up with the same word, you’d get some points that were essentially meaningless, but gave it a competitive gaming style edge. I played with it for an hour or so before boredom set in.
I don’t think this on it’s own is going to lead to Google dreaming in pictures, but I do believe they’re trying.