Originally posted by rezial666
reply to post by America Jones
Good call. But that was 1966 and we have progressed much since then. What if this is an alien AI or program? Maybe you have input on my previous
question:
...
Would then an AI program make a grammatical mistake such as below or would it have programmed such mistakes to increase acceptance by it's biological
counterparts?
It's a good point you make, and one I've spent some time thinking about (hence my comment about Sam's response to poetry).
My initial thinking on the matter arose from a consideration of NSA intercept technology at private telecommunications facilities, and the possibility
that keyword filters might INTRODUCE typos to prevent certain writings from showing up in search engines (as a difficult-to-detect form of
censorship).
I'm certainly aware that AI programs have gotten more sophisticated since ELIZA, but in general, AI is designed to identify as relevant the same
sorts of patterns that human cognition would identify as relevant (what use would AI be to human cognition otherise? If one were to argue that an AI
might be designed to identify novel patterns that humans don't otherwise detect, one would still have to account for how and why such a novel AI
would be useful to humans. If one were to argue that an AI were designed to detect patterns that humans CAN'T detect, what could anyone possibly do
with such information?). My point was more that a familiarity with the most simple forms of AI pattern recognition might help people identify more
complex forms of AI pattern recognition.
Insofar as novel AI is concerned, my primary concern is less that programmers might choose to include tricks (like deliberate typos) to deceive
people, but rather, that people have fundamentally mistaken intuitions about what it is that computers do. People are taught to think of computers as
fancy calculators, but computers are just as much logic machines as they are arithmetic machines.
The most basic operations performed by computer hardware are not mathematical, but consist of precisely one logical operation, called NAND
mathworld.wolfram.com...
NAND is the opposite of the logical function AND. The plain-language equivalent is "at least one is false." It is not taught in most college logic
courses. I am concerned that people are systematically denied basic intuitions about what computers really do, so that people CANNOT suspect what
purposes computers can be made to serve. Washington is colonizing the academies.
Most people are familiar with all the research dollars spent developing fancy equations for rendering 3D models more realistically (especially in the
context of video games), but few people talk about what sorts of novel computing strategies might be constructed around the purely logical aspects of
digital computers.
Because people assume that computers are most useful arithmetically, people have certain intuitions about what communications intercept capabilities
are. People are trained in a million ways to think quantitatively, and people are taught that what is qualitative is subjective, and therefore not a
suitable subject for science.
Quantum mechanics is still seen as exotic, but it's already 100 years old. There's been a lot of research done since then that people are not
taught anywhere unless they're pursuing a PHD in cognitive science, in which case they've read so much jargon-laden nonsense there's practically
nobody they can talk to about whatever it is they think they know.
Scientists have done a lot of work to make information a subject of concrete inquiry. The notion of Kolmogorov complexity, the importance of Claude
Shannon's work for relating thermodynamic entropy to information content, Turing completenes and Godel's incompleteness theorem are strange and
exotic to most people, despite the fact that millions upon millions of people make use of these intellectual innovations every time they sit down at a
computer. These days, 1 bit of information content can be defined precisely as 4 Planck areas (1 Planck area being 10 to the -66 square centimeters).
The universe is information.
Personally, I'm more concerned about living populations being used to perform sophisticated computations, Manhattan Project style, than I am worried
about robots trying to trick me into thinking they're people.
Anybody been experiencing an unusual increase in synchronicity lately?
[edit on 16-11-2007 by America Jones]