it's very much the story of The Solipsist by Frederic Brown, which was published in 1954
Isaac Asimov: The Last Question (1956)
251–260 of 321 posts
Re: Isaac Asimov: The Last Question (1956)
#252It has similarities to a very, very short story by Fredric Brown published two years before. It was called 'Answer' and is only 252 words long: https://www.roma1.infn.it/~anzel/answer.html
yes, i came to look for this comment - i immediately thought of this (i thought of The Solipsist)
Re: Isaac Asimov: The Last Question (1956)
#253[dead]
Re: Isaac Asimov: The Last Question (1956)
#254Earlier quoted context omitted.
As a Trekkie, fair point. I think Star Trek does fall into the "speculative fiction" category, but Star Wars doesn't. It's just "space opera".
Space opera is still a subset of speculative fiction and science fiction, saying "just" dismisses its influence on the genre as a whole. A lot of classic science fiction is basically "x with spaceships" where x is the Napoleonic Wars, or feudal Europe or the Wild West or what have you, and the "science" is little more than set dressing.
> saying "just" dismisses its influence on the genre as a whole.
Well, it was meant to be parsed as:
Star Trek is speculative fiction and space opera.
Star Wars is just space opera.
Some space opera is also speculative fiction, but I wouldn't say it is a subset. I wouldn't call some space opera stories speculative fiction at all.
They're all classified as science fiction.
(Yes, yes - there is no consensus on these terms...typically science fiction is considered a subset of speculative fiction, and here I inverted a lot of things).
Re: Isaac Asimov: The Last Question (1956)
#255>INSUFFICIENT DATA FOR MEANINGFUL ANSWER Boy, it sure would be nice if real LLMs were capable of giving an answer like that.
At the time of this writing, the prevailing thinking with "artificial intelligence" was that we'd encode every Fact we know and every rule of Logic, and from there, the computer would make new discoveries. Todays AI researchers would call this "symbolic" AI, compared to the "neural" AI powering LLMs. They're like two different worlds.
LLMs are just generating text, they don't know anything. They can't assess whether there is enough data for an answer. When you add a follow up prompt "This is wrong, why did you lie?" only then is it able to generate text, "I was wrong, I'm sorry," and so forth.
Re: Isaac Asimov: The Last Question (1956)
#256Earlier quoted context omitted.
if your car doesn' turn left when you turn the steering wheel left, the problem is that the car is broken, if an LLM does something unexpected after you gave it instructions, that's possible when the LLM is functioning entirely correctly.
Nothing in this world is guaranteed. That doesn't mean it's uniformly random either. LLMs can still do something unexpected if you give them clear instructions, but that doesn't mean it'll be arbitrary and unpredictable in scope. The same way C/C++ undefined behavior technically means program can give you nasal demons, but in reality it won't do anything unusual (like format your C:/ drive) unless someone purposefull…
This is all going to flash through your mind when your car mysteriously doesn't turn left. I would prefer to think of machines as things with defined outputs and failure is failure, more than as fluffy little kittens who might do the wrong thing, if the consequences are going to fall on someone who doesn't deserve it.
Re: Isaac Asimov: The Last Question (1956)
#257>INSUFFICIENT DATA FOR MEANINGFUL ANSWER Boy, it sure would be nice if real LLMs were capable of giving an answer like that.
At the time of this writing, the prevailing thinking with "artificial intelligence" was that we'd encode every Fact we know and every rule of Logic, and from there, the computer would make new discoveries. Todays AI researchers would call this "symbolic" AI, compared to the "neural" AI powering LLMs. They're like two different worlds. LLMs are just generating text, they don't know anything. They can't assess whether…
Did Asimov’s idea of AI revolve around data retrieval? I’ve read that even human intelligence isn’t necessarily remembering things, but being able to traverse our knowledge and find that idea or thought quickly.
Re: Isaac Asimov: The Last Question (1956)
#258Earlier quoted context omitted.
The thing is that they are completely incapable of meta-cognition. Reasoning models don’t show their actual reasoning at all.
Right — they're not reasoning, they're generating text that statistically models reasoning. Anyone who says differently is selling something.
Language has reasoning encoded within it.
Re: Isaac Asimov: The Last Question (1956)
#259in the same vein https://www.youtube.com/watch?v=uD4izuDMUQA
Re: Isaac Asimov: The Last Question (1956)
#260>INSUFFICIENT DATA FOR MEANINGFUL ANSWER Boy, it sure would be nice if real LLMs were capable of giving an answer like that.
At the time of this writing, the prevailing thinking with "artificial intelligence" was that we'd encode every Fact we know and every rule of Logic, and from there, the computer would make new discoveries. Todays AI researchers would call this "symbolic" AI, compared to the "neural" AI powering LLMs. They're like two different worlds. LLMs are just generating text, they don't know anything. They can't assess whether…
They can read context and with fairly high accuracy say whether that context contains enough information to answer a posed question. They cannot (and we cannot for them) introspect their own weights to say whether their weights already encode information sufficient to answer a posed question.