Apparently you don't understand how well the right wing can disseminate propaganda. Part of convincing citizens that lowering tax burden and removing regulation is a good idea is convincing them these things are hurting businesses enough to damage the US economy. Telling people that even the anti-business EU has countries which beat us in company formation is a great way to do that. It works even better when people believe a right wing think tank would have some natural bias against giving credit to Europe, so the situation must be even more dire.

Approximating PI = 3 is perfectly valid. But approximating PI = 3.0 is not.

Any time you have a situation where you have to give an answer when a denominator becomes zero is an example of error handling. Just because you have found a specialized way to handle the error while continuing on with your algorithm does not mean the function actually had a value when it divided by zero.

Treating 5/0 as NaN and having your algorithm either skip the value or treat it as zero when computing an average would be an example of handling the error. But it still was an error, although in your specific instance you have determined handling the error was better than fixing your algorithm (which may become needlessly more complicated in your situation).

It has no value. What you have here is a discontinuous function. The graph may appear continuous to the naked eye, but it is not. Just pretending that y = sin(0)/0 = 1 is not proper math.

Nature has no problem with this situation because no practical application of sin(x)/x could ever have x = 0. This is the difference between just computing functions and actually applying them.

It looks like someone just relied on the power rule to get through Calculus 101, and never really understood how what limits are. You use a lot of the terminology someone who understands Calculus would use, but there is a strong disconnect in understanding if you think finding where a function is approaching before it becomes undefined is the same thing as defining an undefined value.

You do not divide by zero in Calculus, you determine what value the function approaches as the denominator approaches zero.

It is CS 101 stuff, which is why so many people are saying it isn't possible. What I don't understand is how many people are actually defending the OP and agreeing that they would want the same thing. I guess these are the same programmers who cannot pass the Fizzbuzz test.

We understand what the OP was asking for and why it would be handy. We are simply explaining why it is not possible. Not just why it is probably not possible, but why it is literally not possible.

I would also love a programming language which always knew what I wanted it to do during runtime errors automatically; I simply don't see it happening before we figure out general artificial intelligence.

You are correct there. A more accurate statement on my part should have been:

If your code ever has the expression X / X and X is capable of being 0, if your code does not treat this as an error and handle it properly then you have a bug in your algorithm.

Treating 0/0 as NaN is an example of handling the error properly, considering you are no longer treating the result as a valid number.

If your code ever has the expression X / X and X is capable of being 0, you have a bug in your algorithm. It really is as simple as that.

Dividing something into zero pieces is not the same as dividing by zero. If you have no money left after dividing five bucks to five people, then the next attempt at dividing would be 0 / 5, not 0 / 0.

Plenty of lazy programmers or those forced into using frameworks developed by other lazy programmers may be used to this. Perhaps even programmers forced to use primitive languages with no other options are used to it. But anyone else who knows what they are doing do not do this. This is similar to all of those developers who use NULL as if it means 0. Just because a lot of people do it doesn't make it any less idiotic.

Even low level languages have the capability of handling money correctly. It just requires more effort by the programmer, just like anything else done in a low level language. Considering all high level languages eventually have their code compiled down to the lowest level language understood by the CPU, low level languages are certainly capable of any calculation that a high level language is.

There are no specifics in that post. Just some vague mention of semantic vs syntactic content and a claim that it is so obvious you don't even need to explain it.

All I can make from your semantic/syntactic analogy is that the physiology of a human brain and its computational ability may equate to syntactic content, but with no inherent semantic meaning. The consciousness that is created by these computations is what provides the semantic meaning. aka I think therefore I am.

But there is no absolute reason why a manufactured machine with similar computational ability as the human brain couldn't also create a consciousness that could give itself the semantic meaning you are referencing.

You are correct that you can clearly demonstrate that clearing 5 lines simultaneously in Tetris is impossible. But that is because you can be very specific as to why: the longest piece is only 4 lines long. Notice I didn't say something vague like: it is logically impossible to clear 5 lines. I was incredibly specific.

If your answer to why computational approaches to AGI is logically impossible are not also as specific, then you are just spouting nonsense. Just saying it is impossible is not an acceptable answer, and neither are other deflections such as calling it silly nonsense or pure fantasy.

If you don't have specific reasons why it is impossible, which show a thorough understanding of why neuron interaction cannot be replicated by a Turing complete computation system, or why simulating neuron interaction is not enough to create AGI, then just stop replying.

You can never state that something is not sufficient just because no one has figured it out yet. Most likely 100% of all significant technological advances were preceded by people who tried and failed. Often the people who finally figure it out were among the people who had failed multiple times in the past.

To say something isn't possible you have to know a great deal about why it is not possible. For instance we know that silicon transistors cannot shrink indefinitely because of our knowledge of physical limitations such as the width of a silicon atom. We don't say that transistors cannot be 5 nm wide just because all attempts at making 5 nm wide transistors have failed; that would be stupid.

It could not possibly be established that computation alone cannot create general AI because we have not even approached computers with enough transistors to match the 100 trillion synapses in an average adult. Even once we reach that milestone it will take a large amount of research before people could claim it simply isn't possible, and even then there would be a good chance they are still wrong.

Realistically the only way to say it is impossible to create general AI with computation alone is to actually create a general AI by other means. Only then could we possibly understand the process enough to know for sure it couldn't be done another way. Although even in that case there is no guarantee someone won't come up with a more elegant and/or efficient way in the future which only uses computation.

Here is an update to this ruling found in another article:

What else other than computation is required? What is it you think your brain's neurons are doing that cannot be replicated?