October 19th, 2006

ozarque figure

Recommended link; linguistics; Pinker's review of Lakoff....

Steven Pinker's less-than-enthusiastic review of George Lakoff's new book -- Whose Freedom?: The Battle Over America's Most Important Idea -- titled "Block That Metaphor!", is online at
http://www.powells.com/tnr/review/2006_10_19 . Here's the opening paragraph as a sample:

"The field of linguistics has exported a number of big ideas to the world. They include the evolution of languages as an inspiration to Darwin for the evolution of species; the analysis of contrasting sounds as an inspiration for structuralism in literary theory and anthropology; the Whorfian hypothesis that language shapes thought; and Chomsky's theory of deep structure and universal grammar. Even by these standards, George Lakoff's theory of conceptual metaphor is a lollapalooza. If Lakoff is right, his theory can do everything from overturning millennia of misguided thinking in the Western intellectual tradition to putting a Democrat in the White House."
ozarque figure

Enlightening and instructive quote....

Quoted this morning at http://www.andrewtobias.com:

"Last year The Boston Globe offered an illuminating comparison: when Bill Clinton was president, the House took 140 hours of sworn testimony into whether Mr. Clinton had used the White House Christmas list to identify possible Democratic donors. But in 2004 and 2005, a House committee took only 12 hours of testimony on the abuses at Abu Ghraib." 
– Paul Krugman, The New York Times
ozarque figure

Writing science fiction; question for the e-savvy/AI-savvy/?-savvy among you...

This isn't going to be a post, because I don't know what I'm talking about here -- it's just a question.

In Moriarty's Spin State and Spin Control, there's a major character -- named Cohen -- who is the artificial intelligence equivalent of a Superhero (or maybe a Supervillain). At least 200 years old, able to assume the bodies of human beings at will, definitely post-Singularity, equipped with incomprehensibly vast neural-net powers, and so on and so on and so superlatively on. Telepathic. Able to carry out feats that look miraculous to the human and altered-human eye. The super-AI to end all super-AIs, although he/she/it started out originally -- I think -- as a human being.

So far, so good. The thing is, when Cohen talks, he/she/it talks just like anybody else.

[I'm not referring to the fact that the English spoken in this obviously Far Future fictional universe is essentially identical to the English spoken in 2006 except for a handful of techno-neologisms; that's a separate issue. Presumably that happens because Moriarty wants to get on with the story, the same way Star Trek throws in the "universal translator" widget in order to get on with the story. Improbable as that is scientifically, to a linguist, I can understand why it's done and what lies behind it; that's clear to me.]

What I'm wondering is whether it's plausible that so advanced an Artificial Intelligence would talk just like everybody else in the novels talks. For example, would the ability to speak (and understand) language indistinguishable from the language used by others over the course of 200 years be something that could just be part of the set of skills programmed into Cohen? Part of his/her/its software, or hardware, or wetware, or whatever term would be applicable?

That's not my only question on this issue, but I don't even have the vocabulary to ask the others; they're just floating around formlessly in my head.
ozarque figure

Writing science fiction; "super-AI" question; afternote...

Thank you for all your responses; they've been very helpful.

I think about how much difficulty an extremely intelligent and hyper-educated and hyper-informed human being -- X -- has carrying on a satisfactory conversation with another human being who is of average intelligence and uneducated and uninformed -- Y. How hard it is for X not to come across to Y as arrogant and patronizing, or worse. How hard it is for X to perceive Y's intentions and emotions and motivations accurately, and vice versa. And then I think of how much greater the distance would be in terms of intelligence and education and information between a "super-AI" like Cohen and a human being -- even an "augmented" human being.

Still, your comments tell me it's plausible. That's baffling, but it's certainly interesting.