science with Lem

[Miri]

How scientific are Lem´s texts?

I am writing a book about science in polish and russian science fiction. What about Lem´s scientific ideas? Are they innovative, are they accurate...? Any ideas?

For example: What about the idea  of a machine evolution in the book about the ship investigating the disappearance of its predecessor?

Or what about the time loops in Pirx star diaries?

[Deckard]

One of the greatest science-fiction ideas Lem depicted in his novels was for sure Golem XIV. He described there a supercomputer outfitted with AI algorithm far exceeding our intelligence. The language Golem XIV uses is very strange and complicated, but of course it was intended and was to emphasize his superiority.  
Of course AI hasn't been yet discovered, so Lem's idea is still valid. This is absolutely the best book on artificial intelligence I read, and since I didn't read any other on AI, it must be the best.... Hold on! I just recalled HALL 9000 was very nicely described in 2001 (by A.C. Clarke). Nevertheless Golem XIV is a lot better, 'coz AI wasn't the main plot of Clarke's novel.

OK, but seriously Lem's novel is really good. The author shows the supercomputer to be nearly a God to us, because it's so unique and smart. On the other hand, despite his divine features, Golem XIV had a problem talking to us, 'coz his perception, intelligence, general and detailed knowledge were so radically better then ours, then when he was talking to us, it was for him like talking to kids. I like the part of the book where Golem XIV is trying to explain why talking to us (humans) is so problematic. One of his dreams was to get in touch with another AI computer - Honest Annie (he finally got his chance). Annie was considered to be the only computer on a similar level of development that Golem was.

Lem presumed in his novel, that modern computers would utilize light. We know now, that future computers will be the photonic ones. There are already some prototypes but the exact date of a massive production has not been yet stated. So Lem was right, that photonic computers will be the most efficient ones and they probably be used as first AI machines.

CU
Deck

[innate]

Machine evolution: Shifting toward miniaturization and decentralization seems likely enough. When we design things, we use the principles that are important for large-scale terrestrial visual animals (e.g., we keep gravity in mind, but generally not, say, surface tension). Evolution will discover many, many niches to fill.

Golem XIV: I should reread that. I remember disagreeing with some of Golem's claims about human evolution, but I can't remember the details.

Incidentally, Lem seems particularly proud of his predictions on "phantomatics" (virtual reality).

[Miri]

I´ll have to check on Golem. Don´t know it yet.

Concerning the machine evolution: am no expert on evolution ideas, but is it probable that those mini-machines (great idea that they unite to bigger units when necessary and fall apart as soon as they are in danger - that means they can become any machine they like) kill humans by erasing their memories? The scientists in the story guess (they don´t really know) that those mini-units empty human minds to make them helpless and thus kill them. But are those humans a threat to their niches? What are those niches anyway? Food = energy, what do those machines consume they could defend against others? And why not kill humans in an efficient way instead of making them stumble over the planet until they die?
And: can´t those machines be defeated because they are so superior in the given evironment or is it, that human knowledge and science is not capable of understanding the perfectly different? The logic of another world? Those questions interest me. Lem really makes me think.

He was very unsatisfied with his own astronautic ideas in Astronauci as they were inaccurate even for the knowledge of his time. Fascinating.

[Miranda]

Where can I find the fantomatics? Started reading all Lems, but have still some more to go...

AI in Golem: So Lem was visionary with light communication in computers. What about the idea of a superior computer? Any signs for such a hugh omniscient computer compared to that people are mere children?

[Deckard]

Fantomatics was widely disscussed in "Bomba Megabitowa" (eng. "Megabit Bomb"), but I was told it wasn't translated into English.

CU
Deck

[innate]

mini-machines: I haven't read the story in some time, but I think Lem will sacrifice scientific rigor in detail in order to say "that which is alien is completely unknowable by man" and "when man does not understand something, he falls back on his primate nature and starts smashing things blindly".

huge omniscient computer: The Singularitarians expect it, but I don't know about the likelihood of it actually happening in the reasonably near future.

[Deckard]

The truth is that the problem is the AI itself. It hasn't been discovered yet. So it doesn't matter how fast and efficient the computers gonna be in future. Without algorithm they will still be as dumb as they are today.

CU
Deck

[Terminus]

Yeah, super efficiently dumb

As a known AI sceptic, what can I say

[innate]

AI algorithms: Do you think the best course of action is to borrow liberally from neuroscience or to try to do something de novo?

[Deckard]

AI is a virgin subject. Many people had already tried to find out the right solution and as we know we didn't have any strict answer. Neuroscience can be very helpful for IT scientists working on the AI. Actually, it's not only about neuroscience - every new discovery in biology, medicine, mathematics, physics, electronics, robotics... you name it, can be useful for the AI research.

CU
Deck

[Terminus]

AI algorithms: Do you think the best course of action is to borrow liberally from neuroscience or to try to do something de novo?

De novo ab ovo.



But seriously: even slowest progress can somehat converge; limit, however, may never be archieved.

[Socrates]

http://www.theonion.com/content/node/30291

hehehehe...

truth or myth?

Cheers, Socrates

[Deckard]

Is it April 1st?
That's a really good joke...  

CU
Deck

[Terminus]

Gosh, what a rubbish...
Didn't buy that, though...