/atheism/ - Atheism

>>11971

Maybe computers can work around it. There are ways to write a software that can look for things that are too complex for our mind but still work.

>>11979

You realize reasoning is ultimately computation right?

http://link.springer.com/article/10.1007%2Fs10994-013-5335-x

>>11979

I don't talk about classic AIs. (But I don't think they are impossible )

No they don't use reasoning but they don't have to. It's aready possible to write software that learns . Give them many large texts and learn if a new text is real or gibberish. The software doesn't know the context but "get" the grammar. I can't descibe it well but google made an algorithm than can detect cat videos just by letting the software watch some videos.

>>11993

The problem is these computers can't cumunicate their results to us. They just show us the result and it's usually right. We get dependent.

>>11995

I had teacher say that to about pocket calculations back when I was a kid.

I think they were right. If I worked out those formulas with paper and pencil I bet I'd still remember them.

Rote memorization and drills are boring but they do make you learn.

>>11968

>Is human knowledge limited?

is our current knowledge lacking? Of course.

Will there be questions evading us forever?

maybe, but none really knows.

In computability theory any problem or question can be converted to an equivalent yes/no question, and the information asked by the original question (like which, how many, how much) is naturally laid out as the consequences of the process of searching for the equivalent yes/no answer. Many of those yes/no questions are in principle always solvable given enough time and enough space to perform the process that leads to an answer. However there are some other questions that you can pose only by modelling the halting of the process to mean say yes, and the perpetual repetition of the process to mean the opposite. However it is impossible to know in advance whether an arbitrary answering process will ever halt or will keep running forever, so those kind of yes/no questions are only solvable when the answer is "yes, it stops", but not when the answer is no. When the answer is the opposite to what is knowable in principle then it is provable that you will never be able to know it. These kind of questions/problems are called "partially decidable/computable/solvable/calculable/answerable/recursive/enumerable"; as opposed to the first kind of questions I mentioned which are simply called "decidable/computable/solvable/calculable/answerable/recursive/enumerable," or "completely decidable/computable/solvable/calculable/answerable/recursive/enumerable".

Note that there are questions which are completely unsolvable no matter what the yet-unknown yes/no answer actually is. One such example is the aforementioned halting problem. I think we all agree that the process of answering a question either stops or runs forever. This shows that there are questions with a valid yes/no answer for which not even one of the two possible outcomes can be known, if you grant me that the full undecidability of the halting problem has been already shown. And all this baggage presupposes that we are dealing with sensical questions for which asking an answer is meaningful. Keep in mind that questions for which no answer in either sense is either knowable or unknowable but simply nonexistent also exist. Questions like "did (the husband who never beat his wife) keep beating his wife?", or "did the celibate keep beating his wife?", or "what is the color of jelousy?". There's nothing to know there, no need to worry about undecidability because the answers do not exist at all. However, proving that a question is meaningless is a separate issue, also subject to undecidability.

I think many of the questions for which "god" or "religion" or "soul" are popular answers are questions which not only do not provide the correct justification to actually know or scientifically suspect with a good degree of certainty that those are the actual answers, but that they also are partially unknowable (the scientific equivalent being unfalsifiability) or fully unknowable; or that they are pseudo-answers to outright nonsensical questions, or that fail to show that they are meaningful questions deserving an answer.

Now imagine the consequences if OP's epistemological question were, at the very least, like a recursively enumerable problem/partially recursive, or in other words requiring no less than a Turing-complete system to be solved. That is the second class of questions that I mentioned above (only answerable in one way). Doesn't sound as bad as a fully undecidable or nonsensical question right? Think again:

I can't conceive of a method that could put that question to test, but it might be possible that one day we will somehow convince ourselves that we are at least in principle capable of knowing anything knowable (being able to know anything doesn't mean it would be feasible to know everything, bear in mind), or the other way around, that humans or science or whatever is ultimately limited. But if we had the right method to know the affirmative case, as long as no answer comes we have to deal with the possibility of facing a question that might be impossible to solve for the negative case, which would mean the philosophical parade keeps running forever, gratuitously, in futility. If the answer exists and it is "no" then the very question we are asking by construction is one of those questions that make the answer what it is. The question you are asking OP would be an example to prove that human knowledge is limited, yet we wouldn't be able to know that we are facing the very answer by asking the question.

You can find on youtube a great conversation between Lawrence Krauss, Daniel Dennet and Massimo Pigliucci about the limits of knowledge and science

>>11970

on the bright side we also live more and have more time to learn things, better methods, more effective learning tools, more access to knowledge. I genius today could learn faster and easier than a genius of the past. He still wouldn't be able to be a polymath as encompassing as the polymaths of the past though.

>>11971

>Could humanity hit an intellectual wall and be faced with problems that aren't just beyond the mental abilities of the average person, but entirely beyond human comprehension?

could the problems we can solve be so numerous that we will be busy solving them forever?

>>11974

>Organisms use the minimum of resources required to survive.

not exactly. evolution does not necessarily produce optimal results, only the ones that are good enough

>>11976

>Maybe computers can work around it

computers could enhance abilities we already have, even by great margins in areas like reliable memory and number crunching, but not provide us with new abilities by mere addition of more and more computer hardware as we know it.

>There are ways to write a software that can look for things that are too complex for our mind but still work.

I agree

>>11979

>Unless I'm wrong.. the machines don't really think

this is an open question. the problem is that we don't know what the nature of thinking really is.

Well, let me back off. I think there are good evidential reasons to believe that human reasoning is a natural process and that computation better describes it, like >>11985 is saying. We have actually past the point of absolute ignorance about the human mind, and even if we didn't I still wouldn't buy into the empty claims of souls and dualist mumbo jumbo. The issue is that there are many things about the nature of the mind that still need to be figured out.

You are right about one thing. Our computing machines and biological brains usually work in very different ways. But that doesn't mean machines cannot think; just as the quite obvious differences between birds and airplanes don't exclude the latter from flying.

>But I don't think a machine use deductive or inductive reasoning

In fact the formalized version of deductive reasoning we call mathematics is the very basis that makes computers work, and it is good enough to model inductive reasoning too.

There are more interesting questions about the abilities of computers and how they compare to the human mind.

>It merely does things that we tell it to

yes. so what?

could it be possible to program them in such a way that they can readapt themselves to deal with all the problems and environments we deal with? could we program them in such a way that they are able to determine themselves resembling the spookiness of free will?

what if we are like that, but programmed by evolution to grow a brain structure that is capable of going without further repurposing other than the environmental inputs and the brain own capabilities?

>>11994

interestingly, all the wave of statistical techniques in AI like machine learning are inspired by how we learn to think. humans go through an incredible amount of training too, so it is unreasonable to ask for an AI that is completely symbolic and laid out from the very beginning to deal with all edge-cases of real life. Still there's a way to go before we have a general AI