The approaches about natural language programming

[QuantumRobin]

The approaches about natural language programming are describe here...

Approach #1: Brute Force Crowd Source. This is the method used in Amazon's ALEXA, Apple's SIRI, Wolfram's ALPHA, Microsoft's CORTANA, Google's HOME, etc. In all these cases, a programmer imagines a question or command that a user will give the machine, and then he writes specific code to answer that specific question ("Alexa, what is the temperature outside?") or carry out that particular command ("Alexa, turn on the living room lights"). Get enough imaginative programmers to write enough routines, et voila! Apparently Intelligent machines that actually exist and work and learn and grow, today.

Approach #2: Dynamically-Generated-User-Tweaked code. This is essentially describe here...

If the programmer is happy with the generated code, (s)he can approve of it and it needn't be saved because it will generate correctly each time before compiling - a label would be attached to the high-level NLP program to tell the compiler that it compiles correctly. If the generated code isn't right though (or isn't complete), that label will not be attached to the NLP code and the support code will need to be saved as part of the program instead. Some of that support code could still be auto-generated initially, creating the loop and setting up the count, for example, while leaving the programmer to fill in the content of the loop manually.

Approach #3 is the one where you build AGI first so that it can solve all the programming problems itself.

What are the programmers' statements about the approaches I quoted above?

Maybe the problem with the Approach #2 is that we need a little more detail regarding the middle step:

Note that I'm not saying Approach #2 is a bad idea or a pipe dream; all I'm saying is that maybe there is not a small prototype based on this approach that can be scaled up to the real deal.

Approach #2 is an optional intermediate step towards approach #3. Approach #3 is the one where you build AGI (artificial general intelligence) first so that it can solve all the programming problems itself. The idea is that instead of the human writing the difficult bits of code to complete a program, the human teaches the AGI (artificial general intelligence) how to write the difficult bits of code so that it won't need help with the same kind of problem the next time. It's all about giving the AGI (artificial general intelligence) system more and more problem-solving skills until it can do as good a job as the best human programmers.

Good to see that at least some good came of this thread. Thanks Schol-R-LEA.

@Schol-R-LEA,

Please talk about the Approach #1 (Brute Force Crowd Source), Approach #2 (Dynamically-Generated-User-Tweaked code) and Approach #3 that I quoted above.

Will you talk about Approach #1 (Brute Force Crowd Source), Approach #2 (Dynamically-Generated-User-Tweaked code) and Approach #3 that I quoted above?

[QuantumRobin]

@Schol-R-LEA,

I edited the last message that I posted in this topic.

@Schol-R-LEA,

Again:

Please talk about the Approach #1 (Brute Force Crowd Source), Approach #2 (Dynamically-Generated-User-Tweaked code) and Approach #3 that I quoted above.

Will you talk about Approach #1 (Brute Force Crowd Source), Approach #2 (Dynamically-Generated-User-Tweaked code) and Approach #3 that I quoted above?

[Schol-R-LEA]

It's so clear to me now - QR is a test of DavidCooper's AGI! Sorry, you aren't passing a Turing Test yet.

[QuantumRobin]

It's so clear to me now - QR is a test of DavidCooper's AGI! Sorry, you aren't passing a Turing Test yet.

@Schol-R-LEA,

I am not a test of DavidCooper's AGI!

@David Cooper,

Please try to show to Schol-R-LEA

that I am not a test of your AGI!

[Schol-R-LEA]

Sorry, that was a rather catty comment, I apologize.

That having been said, I do think you'd find more practical help in a forum which is actually about NLP, rather than a topic such as OS development which is almost, but not quite, completely unlike NLP.

Honestly, I've always been puzzled by your posting here, since you don't seem to have any interest in operating system development, even going back to your original account (Trident). It comes across as if you grabbed on to this forum at random and decided to stay regardless of whether you have any interest in the topic or not, which makes no sense to me. I can understand why DavidCooper is here, as he is working on an OS as part of his larger work; but so far, you don't seem to discuss the topic at all. Wouldn't a different venue - one closer to your interests - make more sense?

I really don't have much to say about natural language processing in general, as I have never really looking into the topic in depth. I have looked up some relevant videos, though, which will at least give some context about previous art on the topic:

• Crash Course Computing: Natural Language Processing
• How complex is natural language? The Chomsky Hierarchy
• AI Language Models and Transformers
• Computerphile: Same Story, Different Notation (it's about the equivalence of finite state machines and regular grammars, but it relates to the issues of the Chomsky Hierarchy, as well as to lexical analysis of regular expressions as it is used in things like compilers and regex engines, though it is only tangentially relevant to NLP, really)

I doubt that any of this is new to DavidCooper, but it might help you, QuantumRobin. I would also recommend reading Gödel, Escher, Bach (or at least reading part of it, no one gets through it all on their first try). If nothing else, the book should do a good job of blowing your mind, just as it did for me.

I will say that part of the problem is that language in humans doesn't build out of any sort of logical structure; it is, shall we say, a hardware pattern recognition function, one which evolved over hundreds of thousands of years, not something that had an sort of goal-directed structure. I would bring up genetic algorithms as a possible direction, though the analogy to natural selection which that is based on is somewhat shaky IMO.

(Again, I expect that David has looked into this at least to some degree before.)

I do hope that both you and DavidCooper took the time to read my (admittedly very long) post about Gödel's theorems, I tried my best to clear up what appeared to be some misunderstandings about it and its relevance to computing.

On a side note apropos of nothing (well, apropos some of the things from Hofstadter's book, but otherwise unconnected to this discussion), the YouTube channel Extra Credits has an excellent series of videos on the development of non-Euclidean geometry and its relevance to modern physics. Just thought I would mention this.

[DavidCooper]

I discussed the Gödel issue with a mathematician a decade ago, starting from the more simple idea that led Gödel to it. This involved a statement with a wording such as "this statement is undecidable" where "undecidable" relates to whether it's true or false. What my analysis of "this statement is true" demonstrates is that that sentence is neither true nor false: it's merely vacuous, there being nothing there capable of holding a truth label. The same applies to "this statement is false", and to "this statement is undecidable" - they all suffer from infinite recursion and never get to anything that can hold a truth label. I expressed a hope that that defect didn't also affect Gödel's incompleteness theorem, but he showed me that the same infinite recursion existed in that and he showed me where it was magicked away with a move which he called the diagonalisation process. The justification for the diagonalisation process came from the precedent set by "this statement is true" being accepted as true, but it isn't true, so the move is an illegal one.

By the way, Quantum Robin puts everything through Google Translate to get it into Portuguese, so none of those video links will be accessible to him.

[QuantumRobin]

Sorry, that was a rather catty comment, I apologize.

That having been said, I do think you'd find more practical help in a forum which is actually about NLP, rather than a topic such as OS development which is almost, but not quite, completely unlike NLP.

@Schol-R-LEA, @David Cooper,

What are the forums which is actually about NLP?

Honestly, I've always been puzzled by your posting here, since you don't seem to have any interest in operating system development, even going back to your original account (Trident). It comes across as if you grabbed on to this forum at random and decided to stay regardless of whether you have any interest in the topic or not, which makes no sense to me. I can understand why DavidCooper is here, as he is working on an OS as part of his larger work; but so far, you don't seem to discuss the topic at all. Wouldn't a different venue - one closer to your interests - make more sense?

Yes, @Schol-R-LEA, a different venue
- one closer to my interests - make sense.

@Schol-R-LEA, @David Cooper,

What are the different venues - the closer to my interests?

Again:

What are the forums which is actually about NLP?

[Schol-R-LEA]

I discussed the Gödel issue with a mathematician a decade ago, starting from the more simple idea that led Gödel to it. This involved a statement with a wording such as "this statement is undecidable" where "undecidable" relates to whether it's true or false. What my analysis of "this statement is true" demonstrates is that that sentence is neither true nor false: it's merely vacuous, there being nothing there capable of holding a truth label. The same applies to "this statement is false", and to "this statement is undecidable" - they all suffer from infinite recursion and never get to anything that can hold a truth label. I expressed a hope that that defect didn't also affect Gödel's incompleteness theorem, but he showed me that the same infinite recursion existed in that and he showed me where it was magicked away with a move which he called the diagonalisation process. The justification for the diagonalisation process came from the precedent set by "this statement is true" being accepted as true, but it isn't true, so the move is an illegal one.

TL;DR: "You [mathematicians] keep using that word. I do not think it means what you think it does."
or,
Layne's law strikes again!

I have the distinct impression that either the mathematician didn't really understand it himself (not uncommon, since most mathematicians these days specialize in one or another branch of math, and axiomatic logic behaves very differently from, say, differential calculus or linear programming), or else explained it rather poorly (possibly both). The whole issue of 'undecidability' in Gödel's paper is regarding the ability to 'decide' a proposition within a given propositional calculus, not in any absolute terms.

I am not entirely sure what you are trying to express here, either, but my guess is that it ties into the terms 'true' and 'false', terms which, unfortunately, don't mean the same thing in axiomatic logic as they do in the everyday world - or in most other branches of mathematics, where they mean yet another thing.

Ordinary use of the word 'true' in English means 'factual', but in mathematics, it only means 'proven' - or more precisely, 'derivable from a given minimum set of axioms'.

It seems reasonable to assume that these were in fact the same thing, but... they aren't. Mathematical 'true' is an analogy to real-world 'true', but by the late 18th century you began to see topics where that analogy didn't quite hold (e.g,. imaginary numbers, higher-dimensional geometry), and mathematicians have been coming a'cropper of that ever since. This disconnect - and similar ones about the meanings of terms such as 'point' and 'line' - was at the heart of several major upsets in modern mathematics, such as non-Euclidean geometry, and unfortunately, it is part of the interpretation of Gödel's proofs as well - which is part of why Hofstadter compared the two sets of developments, as I mentioned yesterday.

Worse, in propositional logic, it doesn't even mean 'proven' - it means 'well-formed according to the rules of this particular logic system'. Part of the fallout of Gödel's proofs was the realization - which you'd have expected people to figure out sooner, given that the idea of well-formed but counterfactual syllogisms was at the core of the earlier Aristotelian logic which symbolic logic arose from, but hindsight is 20/20 - that on the one hand, a given calculus' rules always had some element of arbitrariness to them (because of how axiomatic systems work), while on the other, the domain of derivable rules for a consistent propositional calculus is always a proper subset of (either equal to, or subsumed by) the set of well-formed (== 'true') propositions of that calculus, but for one which is complete, there are 'true' ( == 'well-formed') rules which can be interpreted as contradictory to other 'true' propositions.

I suspect the word 'interpreted' is relevant too, but as I have said, this isn't an area I an on solid ground with. Let's just say that just because something is provable (or, to compare it with quantum physics, demonstrable/factual) doesn't mean everyone will agree on what it means semantically. It ties into the 'well-formed' part because a well-formed proposition is only mathematically 'true' (== 'a proof') if you interpret as corresponding to something in general mathematics.

The problem is that, for a calculus which is consistent, if one 'runs the rule backwards' - that is, creates a well-formed proposition to match the mathematical proof, rather than by mechanically following the rules of the specific propositional calculus - one can find 'true' (== 'well-formed') propositions which nonetheless cannot be found through any sequence of transitions within the rules of that calculus. Conversely, for an inconsistent calculus, you can always find at least one pair of propositions which - when interpreted outside of the calculus as mathematical proofs (== 'true') - contradict each other as statements of mathematical proof.

While in some senses this is just restating what I've already said, I hope that it gives a different enough angle to it that it helps in understanding it.

[QuantumRobin]

Sorry, that was a rather catty comment, I apologize.

That having been said, I do think you'd find more practical help in a forum which is actually about NLP, rather than a topic such as OS development which is almost, but not quite, completely unlike NLP.

@Schol-R-LEA,

Again:

What are the forums which is actually about NLP?

Honestly, I've always been puzzled by your posting here, since you don't seem to have any interest in operating system development, even going back to your original account (Trident). It comes across as if you grabbed on to this forum at random and decided to stay regardless of whether you have any interest in the topic or not, which makes no sense to me. I can understand why DavidCooper is here, as he is working on an OS as part of his larger work; but so far, you don't seem to discuss the topic at all. Wouldn't a different venue - one closer to your interests - make more sense?

@Schol-R-LEA,

Again:

Yes, @Schol-R-LEA, a different venue
- one closer to my interests - make sense.

@Schol-R-LEA,

Again:

What are the different venues - the closer to my interests?

[DavidCooper]

@Schol-R-LEA, @David Cooper,

What are the different venues - the closer to my interests?

Again:

What are the forums which is actually about NLP?

There might not be any forums specifically about it, and particularly if you're referring to natural language programming rather than natural language processing, but it's a topic better suited to a forum about programming in general rather than one that's specifically about operating systems, even if the two things converge in the future (with every OS being built as an AGI system too and with full support for natural language programming).

[QuantumRobin]

@Schol-R-LEA,

You said to me about forum which is actually about NLP in the following response:

Sorry, that was a rather catty comment, I apologize.

That having been said, I do think you'd find more practical help in a forum which is actually about NLP, rather than a topic such as OS development which is almost, but not quite, completely unlike NLP.

Honestly, I've always been puzzled by your posting here, since you don't seem to have any interest in operating system development, even going back to your original account (Trident). It comes across as if you grabbed on to this forum at random and decided to stay regardless of whether you have any interest in the topic or not, which makes no sense to me. I can understand why DavidCooper is here, as he is working on an OS as part of his larger work; but so far, you don't seem to discuss the topic at all. Wouldn't a different venue - one closer to your interests - make more sense?

@Schol-R-LEA,

Again:

Yes, @Schol-R-LEA, a different venue
- one closer to my interests - make sense.

@Schol-R-LEA,

I am waiting your answers on these questions:

Again:

What are the different venues - the closer to my interests?

What are the forums which is actually about NLP?

[DavidCooper]

The whole issue of 'undecidability' in Gödel's paper is regarding the ability to 'decide' a proposition within a given propositional calculus, not in any absolute terms.

If a propositional calculus allows irrational moves to be made, then it's fantasy mathematics rather than anything of relevance to the real world, but if that was the case, it would also mean that people are also making a big mistake when they try to apply it to the real world, which they frequently do.

When we have AGI, we can let it untangle all the mess and judge it for us, putting each kind of maths in the right category so that people can see whether it's real or fake. What actually matters though in this conversation is the simple point that "this statement is true" is not true. I have not heard any mathematicians making that point, and I keep seeing "this statement is false" being presented by some as an unresolved paradox, but I have resolved it. And the point of me bringing this into the conversation here was simply to show something I've found which anyone with a good mind should be able to understand and recognise as correct. It was to illustrate that I have the ability to break new ground and that people should not keep telling me that I can't be doing the work that I know I'm doing. There are other things I could point you to which show the same thing, such as my demolitions of Einstein's STR and GTR . Here's one of my disproofs of STR: https://www.quora.com/What-are-the-stro ... ooper-613?. Again though, people merely dig in to defend the deities of science rather than looking at the places where they've broken fundamental rules. You can't rely on getting recognition for rightness from humans because their thinking is overridden too easily by what they want to be true. Here's another one: http://www.magicschoolbook.com/science/relativity. You can show people as much proof as you like, but what actually happens is that they simply reject it by waving their hands about and saying "it must be wrong", but none of them can break the argument even though I've gone to ridiculous lengths to make it as easy as possible for them with the "interactive exam" which should pin their objection down to a specific point, if they really had one. They are not rational: Einstein has become a religion. I even have a long series of emails exchanged with a physicist in which he ended up rejecting mathematics and asserting the superiority of physics over it.

It's clear now though that the only thing people will accept as a demonstration that I'm the kind of person that can build AGI is for me to build it, but that's fine because that job is practically done. The system exists and it's just a matter of debugging it and then loading it up with knowledge, plus a few tweaks here and there to provide any small essential bits of functionality that might have been overlooked and left out. Quantum Robin has merely forced me to say a few things about it here a few months earlier than I wanted to. I still aim to have a demo ready by the beginning of March, but if you happen to be in Scotland for any reason after the middle of the year and the demo still hasn't materialised, I'd be happy to show you exactly where things are. I can give you a tour of the complex interlinked structures which replicate every single detail of what the brain must also do with its data in order to support all the general intelligence functionality which we know that it has, and I'll show you the extent of the code and explain the algorithms which it applies. Obviously I'd rather not have to show you the internals of it at all and I'm confident that I won't need to, but there you have a promise which you can follow up on if you have to. The project is real and I don't want to discuss it here again until March.

[QuantumRobin]

The whole issue of 'undecidability' in Gödel's paper is regarding the ability to 'decide' a proposition within a given propositional calculus, not in any absolute terms.

Quantum Robin has merely forced me to say a few things about it here a few months earlier than I wanted to. I still aim to have a demo ready by the beginning of March, but if you happen to be in Scotland for any reason after the middle of the year and the demo still hasn't materialised, I'd be happy to show you exactly where things are. I can give you a tour of the complex interlinked structures which replicate every single detail of what the brain must also do with its data in order to support all the general intelligence functionality which we know that it has, and I'll show you the extent of the code and explain the algorithms which it applies. Obviously I'd rather not have to show you the internals of it at all and I'm confident that I won't need to, but there you have a promise which you can follow up on if you have to. The project is real and I don't want to discuss it here again until March.

Hello @David Cooper!

Thanks for your responses!

@David Cooper,

I will send other email to you in 31 March 2020 to know if you created NLP and AGI.

I do not intend to send other email to you before 31 March 2020.

@David Cooper,

What is your opinion about it?

[Schol-R-LEA]

The whole issue of 'undecidability' in Gödel's paper is regarding the ability to 'decide' a proposition within a given propositional calculus, not in any absolute terms.

If a propositional calculus allows irrational moves to be made, then it's fantasy mathematics rather than anything of relevance to the real world

Well, yes, and that was the point - that propositional calculi were a dead end for automatic theorem derivation, because they would all have that problem. The idea of a propositional calculus as a way to find all mathematical theorems by a purely mechanistic process, without also finding false positives, doesn't work. That's all Gödel's theorems really say on their own; they are about the limits to which formalization can be taken.

To put it in a way more familiar to most grammarians (and compiler developers), what they really say is that syntax isn't semantics.

(Assuming you are talking about a 'sufficiently strong' language i.e., one which cannot be described in terms of a regular grammar, so ones requiring either a context-free or a context-sensitive grammar to describe. Regular languages are vacuous enough that they don't really carry any additional semantics unless you deliberately impose an interpretation in them, as you would when using one for, say, a regex parser.)

Do Gödel's theorems apply to any possible mathematics? Not directly, because there is nothing inherent to mathematical reasoning that requires a strictly formal approach. Does it necessarily apply to any mechanical system trying to work out mathematical problems? No, obviously not, since that would mean human brains - which are mechanical processes, so far as we know - couldn't figure out things such as, say, Gödel's theorems. It just means that that particular approach to it, doing it by a fixed set of rules without applying any sort of external interpretation while applying the rules, leads to absurdities. Nothing more, nothing less.

(Assuming you could do that in the first place; as I said, part of what the paper the theorems were published in showed was that most logicians trying that weren't managing to avoid interpretation - indeed, the theorems themselves are based on showing the need for interpretation outside of the calculi.)

It doesn't even mean that propositional calculi don't have value in more constrained uses, just that they aren't a silver bullet for finding mathematical proofs flawlessly.

What bearing does it have on hard AI? No one knows, though honestly, I don't see any reason it would have any at all. But then, I am not convinced (as I joked earlier) that 'intelligence' actually is all that meaningful to begin with (yeah, yeah, cogito ergo sum, but Descartes was pulling a bit of a fast one with that - one which fit his ulterior goal of divorcing philosophy, logic, and mathematics from religious doctrine - by not actually defining what 'esse' or 'cogitere' mean in his system). 'Intelligence' is, metaphorically, a necessary axiom for rational discussion rather than something which can be defined.

I am not convinced that we'd be able to overcome the limitations of our neurological structures when observing a self-aware system that is based on different underlying structures; that is, even if we could create an AI, what guarantee is there that we'd be able to tell that it is an AI, if it thinks in ways which are radically different from the ways we do? The flaws in the Turing Test are pretty well known, and expecting a priori that an AI could even meaningfully communicate with us (or conversely, that a system which appears to converse as we do actually is self-aware) is a pretty shaky assumption. I doubt that it would be possible to prove beyond a reasonable doubt that your AGI is an AGI, even if it is one.

I won't even get into the question of whether creating a new sophont (by any means at all) is ethical or meaningful; suffice it to say that I refuse to have any children of my own.

Mind you, the other part which most people fixate on - undecidability - is not quite what it is usually described to be, either. Basically, it was merely a new tool in the mathematicians' arsenal, a method by which they could reason about reasoning; it showed that you could make proofs about whether a conjecture was decidable (provable or disprovable), by presenting an argument about mathematical logic as a whole as it applied to the topic, even if you couldn't prove or disprove said conjecture directly (i.e., you night not have a way to prove or disprove something, but you could at least figure out if a proof was possible or not). Since mathematicians don't like being frustrated any more than anyone else does, this at least lets them drop fruitless lines of inquiry (e.g., the Halting Problem) ahead of time.

(The other practical effect of proving undecidability of a problem is to say that if you take a conjecture about the truth or falsity of something as an axiom - that is, if you say, "I haven't proven this yet, but here's what it would mean if I had" - and that conjecture is then shown to be undecidable, then the reasoning based on that conjecture is itself undecidable, and thus isn't really usable in mathematics except as a counterfactual; you can do it, but it won't lead anywhere. IOW, undecidability is contagious - any reasoning based on an undecidable proposition can't be used in a decidable proof.)

The point is that it is method of reasoning about mathematics itself, rather than about any particular application of mathematics.

What I am trying to say is the real problem is that Gödel's theorems - what they mean, what they apply to, the supposed profundity of them - has been vastly misrepresented in the popular press, and even the ones who mostly get it right such as Hofstadter are generally misunderstood by those reading their works. So, Inigo Montoya gets the last word after all fnord.

[Qbyte]

What actually matters though in this conversation is the simple point that "this statement is true" is not true.

That resolution has always appealed to me as well, because what would that statement actually be true about? Nothing really, hence it can't satisfy the meaning of truth. A statement is true if what it is saying is correct, that is, it's saying something that is factually correct about something. In that sense, the statement "this statement is true" is practically no different to the statement "this lemon is true". Since neither statement can satisfy the ability to be true, they aren't.

I keep seeing "this statement is false" being presented by some as an unresolved paradox, but I have resolved it.

When resolving this paradox, one should always address the strengthened version: "This statement is not true". So then, what is your resolution of it?