Language, Consciousness and Intelligence?
TAI Archive | Commentaries: "Back in 1887, Alexander Graham Bell shared with the president his view that the telephone was so important that 'someday every community will have one.' In 1889, Western Union president William Orten, rejecting Bell's offer to sell his struggling telephone company to Western Union for $100,000, said, 'What use could this company make of an electrical toy?' In 1943, IBM chief Thomas Watson said, 'I think there is a world market for maybe five computers.' In 1977, Digital Equipment founder and president Kenneth Olson said, 'There is no reason for any individual to have a computer in their home.' In 1981, Microsoft founder and president Bill Gates said, '640K [of memory] ought to be enough for anyone.' And, of course, everyone remembers Charles Duell, the U.S. Commissioner of Patents, who said in 1899, 'Everything that can be invented has been invented.' "
Nathan Myhrvold @ Microsoft (ACM '97): "1) Software is a gas: it expands to fit the size of its container.
2) Software grows until it hits the memory and processing limits of the
current computer technology; it brings old machines to their knees just when
the new model is ready.
3) Software growth drives hardware growth. People buy new models because
their old ones are bogging down.
4) Software is limited only by human ambition and expectations.
- It is often said that we are today in a software crisis. But this has
been said throughout the history of computing. Many software technologies
were supposed to solve it but didn't: high-level languages, object oriented
programming, component software.
- Future software technologies: genetic programming, software husbandry."
The Reality Club: THE SINGULARITY: "Herbert Simon's 1965 prediction that by 1985, 'machines will be capable of doing any work a man can do.'"
Predictions for the Third Millenium (David Krieger): "I think everything that is both presently imaginable and physically possible will be within the reach of human technology within the next century at most, if we don't annihilate ourselves. "
TECHNOLOGY ZOOMS: "Birthing sentient artificial intelligences may be a hell of a lot harder than first thought. The more we findout about the brain, the slipperier the concept of “mind” gets. Even as the mapping of every synapse in the human brain becomes possible, accurate modeling of the function of even a single synapse eludes us. And the mind is not merely the product of the firing of synapses, but of some complex interaction between the brain and body, with everything from culture to the glands having a role to play"
The Tower of Electric Babel: "Language is a guessing game."
Wired News: AI Founder Blasts Modern Research: "'AI has been brain-dead since the 1970s,' said AI guru Marvin Minsky in a recent speech at Boston University."
Scientific American: The World in a Box -- Little Fanfare Greets The Coming Out of a Pivotal AI Project: "[2001] has come and gone with computers eliciting not invigorating repartee but muffled cursing from their users at the obtuseness of their behavior."
"It would have taken a single programmer 500 years to incorporate the almost 1.5 million facts about the everyday world that are in Cyc's database. "
hmm: I'd guess that Dennett and Hofstadter would be happy about the "intelligence" of any computer powered by Cyc; but does it address the combinatorial explosion?
Funding a Revolution -- Thomas Hughes (Nat. Research Council)
Artificial Intelligence
"Support for research in artificial intelligence (AI) over the past three decades has come largely from government agencies, such as the Defense Advanced Research Projects Agency (DARPA), the National Science Foundation (NSF), and the Office of Naval Research (ONR). Firms that initiated AI research programs in the 1960s eliminated or truncated them severely once they realized that commercial applications would lie many years in the future. While not attaining the original vision of creating a truly thinking machine, research in artificial intelligence has generated numerous advances in expert systems, speech recognition, and image processing. Industry is actively commercializing many of these technologies and embedding them into a range of new products. (1999)"
KurzweilAI.net (Douglas Hofstadter): "The number of sentences you'd need to store in order to be able to respond in a normal way to all possible turns that a conversation could take is more than astronomical-it's really unimaginable. And they would have to be so intricately indexed, for retrieval...Anybody who thinks that somehow a program could be rigged up just to pull sentences out of storage like records in a jukebox, and that this program could pass the Turing Test, hasn't thought very hard about it. The funny part is that it is just this kind of unrealizable 'parrot program' that most critics of artificial intelligence cite, when they argue against the concept of the Turing Test."
indexation is trivial.
memory is exploding exponentially.
An inter-lingua based hyper-text document (i.e., 16 bits per item or 65k unique interlocutor-responses and 65k not-necessarily-mapping unique computer responses) at 2 turns per minute requires 20 turns for the "ten minutes" Hofstadter specifies. But let's be unambitious and cover 5 minutes (i.e., 10 turns).
If we "keep" the 100 most probable interlocutor-responses, this requires 100^10~2e20 B of memory.
Given that "petabytes are coming", and that in ten years, 1e15 B will cost $2k (according to Jim Gray), sufficient memory will only cost $400m... and that is only in ten years!
"unimaginable"? We'll forgive DH for his lack of imagination given the date his words were published (1984)...
Jim Gary's Turing talk
"learning and concept formulation are still an elusive goal"
"super-computer has comparable power to human brain"
"human genome is apparently specified in 1MB"
(after stripping away genetic junk and genetically similar material)
"We are missing something:
- great compression algorithm?
- better programming languages?
- learning?"
"long term research projects:
- human-computer interface: make computers easier to communicate with
- organize, summarize and analyze information
- auto-managing & auto-programming & safe
paradoxically, many problems come back to machine intelligence"
Presidential IT Advisory Committee:
Information Technology for the Twenty-First Century Initiative specifies work in:
- Software Research
- Scalable Information Infrastructure
- High-End Computing
- Socioeconomic Research
