h a l f b a k e r yIt might be better to just get another gerbil.
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
| |
We were expecting idea to contain BuckyBalls™. Add them, or perhaps include a geodesic dome. |
|
| |
//too much or too-low-quality content// I don't believe there is such a thing. |
|
| |
No intuition. Just re-hashes of existing content. |
|
| |
The ultimate echo chamber. |
|
| |
The Library of Babel contains every possible text (within certain arbitrary constraints defined by character set and text length). Jonathan Basile's online implementation attempts to obfuscate the completeness by using a clever mapping algorithm to present every page of text in random order so that you do not know which page comes next in order, I suppose to give the thrill of the search. His idea is that the project is deep and profound and raises questions about meaning and significance. |
|
| |
I have long thought that a better way to construct a Library of Babel is to organise the pages in alphabetical order, so that each page differs from the next page only by the final character of that page being incremented by one alphabet position. That way you could instantly find the text you wanted: the room, shelf, volume and page references would be simply the appropriate extract from the text on the page. You lose the thrill of the random hunt but you gain a clarity of the structure of the entire collection. |
|
| |
I suppose one of the points of this kind of complete collection of possible texts is that there is no meaning in the texts. it is pointless saying that "this text I am writing now exists in the Library". The text only "exists" when you go to the library and search for it and find it. The text in the library has no meaning, until the reader finds and reads the text. |
|
| |
I think this is connected to the idea of the "death of the author", where it is the reader of a text who generates the meaning in their own brain. A lot of the argument about this idea seems to center around the obvious problem that the author also has a brain and is thinking about stuff as they write (though one could also analyse the process of writing as involving simultaneous reading-back of the text as it is being written). |
|
| |
An AI written text is really on a technical level not that different from a page in the Library of Babel, in that the machine spits out a text (or selects from a multitude of possible texts) but it only becomes meaningful if and when a reader reads it and creates meaning and understanding from it. The algorithm in the Library of Babel does not seem to create understanding of the texts it curates, it just generates them according to a quasi-random algorithm. My alphabetical library is simpler and the alphabetical algorithm is easier to understand, but Basile's deliberately obfuscating algorithm does not seem to be understanding every possible text either. And so the AI proposed here (the Large Language Models I presume) does not seem to be reading and understanding any of the texts it "generates" (finds from possibility space) either. |
|
| |
Maybe the ultimate conclusion then is that human readers also do not understand anything, it is all an illusion. |
|
| |
I only want to read ideas generated by human minds, that I know are motivated to be creative and delighted/annoyed etc when these ideas are applauded/improved/subjected to feedback of all kinds. I know that ai can imitate these responses but that's the equivalent of fitting a sensor to a table leg so that it yelps when you kick it (piece of my work btw) |
|
| |
//fitting a sensor to a table leg so that it yelps when you kick it (piece of my work// Oh, now I must have the link to this masterpiece! |
|
| |
Finding perfectly meaningful texts in this library by a random pull would be like trying to land on a rational number in the number line just by chance, which is actually zero. But they wouldn't have to be perfect in order to communicate meaning. How much of them would communicate "truth" is an intriguing thought experiment. |
|
| |
That xkcd makes me angry because it doesn't demonstrate a geometrically decreasing quantity between groups of numbers |
|
| |
Or a bakery, where ideas can be kneaded into a kind of absorptive earth to cleanse them from the echoing din of AI. |
|
| |
I'm sure in a few years we'll all just prefer AI. Similar to we prefer web pages over paper. The remnants of humanity that read and write on paper and talk to each other will exist in holes in the ground but will be largely ignored by the megapipes of the streaming on demand economy, where people ingest free and meaningless input and output money obtained by luck, heredity, or hard work. And love it! |
|
| |
What we prefer at any given time is not necessarily what's best for us. |
|
| |
[Voice] do you not pull books apart and only keep the pages that are of interest? |
|
| |
Good idea. Then only allow AIs to comment. Cage fight of the minute, free to watch. No touchee. |
|
| |
Only allow AIs to read. No humans. |
|
| |
//do you not pull books apart and only keep the pages that are of interest?//
Do you do that? How interesting! You must come over for a visit some time, I'll show you my pet shark. |
|
| |
What a shame, What's your address then? I'll send you a very special cup of tea. It's so good we have to mail it in a lead canister to keep the deliciousness in! |
|
| |
Interesting. I'll donate it to the local food bank, they are always soliciting donations of tea |
|
| |
I like the idea of this. Probably more than I should. xen, I think they would make extra of you. Just because. |
|
| |