View source for Artificial Intelligence

Here is a documentary called [https://archive.org/details/DouglasAdams-Hyperland Hyperland] from 1990 by [https://en.wikipedia.org/wiki/Douglas_Adams Douglas Adams], the guy who wrote [http://www.bbc.co.uk/cult/hitchhikers/ The Hitchhiker's Guide to the Galaxy] and some other brilliant, funny, geeky books.

It's about the web, which at the time was still a pretty futuristic idea, but more than just a glint in someone's eye, as you'll find out if you watch it.

It revolves around hyperlinking and a "software agent" (Tom) who was very definitely expected to be '''intelligent'''.  The key phrase for me was when Tom says, "Well, I rather like to arrange meaningful coincidences."  That is what Artificial Intelligence (AI) is about.


::<html5media height="360" width="640">https://archive.org/download/DouglasAdams-Hyperland/DouglasAdams-Hyperland.mp4</html5media>


===Some Thoughts on AI===

For humans, thought doesn't happen in isolation. It happens in the context of languages, like maths and culture etc, which require lots of time and lots of people to develop. Sometimes those languages evolve quickly, mostly they don't, but in a very literal way, super intelligence is what humans already do.

These languages can also limit our thought, so improving them is the way forward for human intelligence. But there's no reason to suppose that we've reached peak efficiency with them - perhaps there's still plenty of headroom, as it were. Personally, I think our ways of sharing things, like the complex thought connections we're able to make as individuals, still have a very long way to go.

While AI might be able to learn things rapidly, and can already outperform human intelligence in specific areas, it's starting from a quite different place, without (for now) the kind of heritage and super-multidimensional connectedness that humans depend on. I'd guess that makes something which would genuinely pass for intelligence (as we understand it) a pretty long way off, even with effective technology.

Ultimately, the sci-fi idea that humans might be undone by their AI assumes they will directly compete. But that hasn't been our historical experience. Agriculture, breeding, machines etc have all deeply effected the status quo for humans, achieving things that humans themselves can't hope to and causing important changes in what humans are all about. I expect AI will be like that.

Quite apart from this, to produce something like meaningful super intelligent AI, our own super intelligence might have to move on. For example, our understanding of how things emerge (like I think intelligence does) is still quite rudimentary.
Related Pages
Collaborative filtering	Knowledge Management	Bottom-up
Related Pages	Machine learning	How to add stuff to a wiki
Tacit Knowledge	Personal Recommendations	Is IT making us stupid?
Filter bubbles	User:Gav	Knowledge web