Comments on: Natural Language Understanding

By: Titus Quinn

Titus Quinn — Mon, 15 Feb 2010 21:18:14 +0000

“Or, for that matter can an AI imagine something that does not exist in reality?”

Depends on how you define “imagine.” I don’t see why not. Really, unless you’re planning on reconstructing the human brain very, very closely, you would be, for all intents and purposes, be creating an alien intelligence (albeit one with which we could communicate and reach a mutual understanding with.)

As such, there’s always some epistemological difficulty with applying words which describe human experience to an alien one. For example, does an AI “see color?” Well, absent the structures of, say, the human eye and optic lobes, it would be hard to say it captures the qualia of color, but certainly an AI could see and discriminate different frequencies of light better than a human could.

By: Titus Quinn

Titus Quinn — Mon, 15 Feb 2010 21:07:51 +0000

There’s no reason to believe hype about AI-created utopia that fails to address the issues of things like natural selection and resource scarcity.

By: Peter

Peter — Sat, 13 Feb 2010 21:01:34 +0000

A.I. is simply a means to accelerate what is our natural evolution.
AI is reality.
We exist in reality ergo !!

Darwins Church the rules do not change.

By: Matt

Matt — Wed, 10 Feb 2010 19:28:59 +0000

As a human, EE Smith can imagine and write about an artificial reality such as the SF story depicted. We can then read his natural language and experience that reality in our own imagination. If it were possible to project each of our interpretations as a movie and compare them side-by-side we would find them all slightly different because each would be colored by our own past experiences.

AI is in fact something that we imagine, another SF story, at least for the moment. All of this makes me wonder how an AI would distinguish reality from fantasy. Or, for that matter can an AI imagine something that does not exist in reality?

By: J. Storrs Hall

J. Storrs Hall — Wed, 10 Feb 2010 15:44:21 +0000

Erin: In the long run, we want something vaguely like that to happen: As AIs get significantly smarter than wild-type humans, it’s only reasonable that more and more of the decision-making should lie in their hands. The pathway to such a state, however, is fraught with difficulty: there will be, indeed already are, many people, organizations, and soon AIs, who want as much power as possible for their own benefit and not for that of the world as a whole. How should we distinguish? In the real world, every attempt to build utopian totalitarianisms have resulted in unmitigated disaster. So I’m highly skeptical about the possibility of such a scheme working well with anything resembling our current state of knowledge.

By: Erin

Erin — Wed, 10 Feb 2010 15:07:44 +0000

Here is an AI question for you, Josh and all:
I was discussing the possibility of human level and super human AI with someone and he says “I would like us to develop an AI “Nanny” that would have control over the entire world’s weapon systems, including any future nanotech and biotech systems, and this AI Nanny would “kindly but firmly” prevent humans from war, murder,rape, attack/assault on one another. Armed with advanced particle beam, nanotech, and other devices our Nanny AI will play the role of Moral Referee for mankind, and at the slightest attempt to harm another intentionally (he says the system would also have access to fMRI and neurological scanning devices), would not allow it.” What are the weaknesses, errors, and dangers with such a “Nanny AI” that this person told me he wants to see happen?

By: Mike Dugan

Mike Dugan — Tue, 09 Feb 2010 15:45:23 +0000

Loved EE Smith.

Nautral?

does anyone spell check?

[Thnx -- fixed -- WordPress spellchecks in the text but apparently not the title! --jsh]

By: Dexter Johnson

Dexter Johnson — Tue, 09 Feb 2010 09:22:35 +0000

Thank you for the mentions. Is Nanodot on Twitter? I would like to retweet your posts.

[Nope, not on Twitter, sorry! --jsh]