Why is The Gettier problem still considered an open issue?

Known unknowns and unknown unknowns!

Moderators: AMod, iMod

PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Why is The Gettier problem still considered an open issue?

Post by PeteOlcott »

Skepdick wrote: Fri Jun 02, 2023 9:23 pm
PeteOlcott wrote: Fri Jun 02, 2023 8:43 pm My purpose in showing that the actual Liar Paradox is simply a semantically incorrect expression of language is to overturn the Tarski Undefinability theorem that utterly depends on the Liar Paradox being self-contradictory.
Take the sentence "This sentence is true" which is most definitely NOT self contradictory.

Code: Select all

?- unify_with_occurs_check(LP, true(LP)).
false.
Ooops!

You fucked up again.
The question that unify_with_occurs_check(LP, true(LP)).
is asking is: Is this expression semantically correct?
Skepdick
Posts: 14347
Joined: Fri Jun 14, 2019 11:16 am

Re: Why is The Gettier problem still considered an open issue?

Post by Skepdick »

PeteOlcott wrote: Fri Jun 02, 2023 10:07 pm The question that unify_with_occurs_check(LP, true(LP)).
is asking is: Is this expression semantically correct?
I know!

Your procedure returns "false" ! e.g it's NOT semantically correct.

Why? What's wrong with it?

There's nothing wrong with it! It's just that unify_with_occurs_check() rejects ALL recursive definitions.
Age
Posts: 20043
Joined: Sun Aug 05, 2018 8:17 am

Re: Why is The Gettier problem still considered an open issue?

Post by Age »

PeteOlcott wrote: Mon May 29, 2023 1:18 pm
Age wrote: Mon May 29, 2023 8:16 am
PeteOlcott wrote: Sun May 28, 2023 4:45 pm In Prolog and the architecture of the formal system that I am proposing expressions are only true if they can be deduced from expressions of language that have been stipulated to be true.
But WHO IS the ONE who gets to STIPULATE what IS TRUE FROM what IS NOT TRUE?

For example, can ANY one 'put their hand up' for that job?
The computer does it itself.
you appear to KEEP MISSING the Fact that a computer can ONLY 'decide' based upon the ACTUAL person, who 'programmed' ' the computer'. A computer does NOT make decisions itself. A computer just 'puts out' based solely upon what has been 'put in', again by 'you', human beings. So, what this means is if there is absolutely ANY bias within the human being, programming the computer, then 'this bias' could all to easily and simply be transferred into 'the computer' as well.

The 'biases', by the way, which ALL of 'you', adult human beings, HAVE can be very easily NOTICED and SEEN.
Age
Posts: 20043
Joined: Sun Aug 05, 2018 8:17 am

Re: Why is The Gettier problem still considered an open issue?

Post by Age »

PeteOlcott wrote: Mon May 29, 2023 1:26 pm
Skepdick wrote: Mon May 29, 2023 1:22 pm
PeteOlcott wrote: Mon May 29, 2023 1:18 pm

The computer does it itself.
And who does the stipulation?
ChatGPT was able to figure out that {baby kittens} are not any type of {ten story office building}
ONLY and SOLELY BECAUSE that information PREXISTED BEFORE that PROGRAMMED 'chatgpt' went and did what 'it' was PROGRAMMED TO DO. Which is to just find ALREADY EXISTING information. 'chatgpt' is NOT programmed to just 'MAKE UP' whatever 'it' DECIDES nor WANTS 'to make up'.
Age
Posts: 20043
Joined: Sun Aug 05, 2018 8:17 am

Re: Why is The Gettier problem still considered an open issue?

Post by Age »

PeteOlcott wrote: Mon May 29, 2023 8:16 pm
Age wrote: Mon May 29, 2023 8:11 am
PeteOlcott wrote: Sun May 28, 2023 2:14 pm

So we can make a Chatbot that unceasingly argues against each and everyone of these people on social media every which way thus quelling counter-factual disinformation before it ever gets started.
But Who is going to program WHICH WAY the chat it is going to LOOK AT and VIEW 'things'?

you keep appearing to completely MISS or MISUNDERSTAND the ACTUAL ISSUE here
PeteOlcott wrote: Sun May 28, 2023 2:14 pm No one will ever make any attempt to do this while they believe that the Tarski undefinability theorem is true.
False.
PeteOlcott wrote: Sun May 28, 2023 2:14 pm I have already shown how prolog screens out the Liar Paradox, and since the Tarski undefinability theorem has only the Liar Paradox as its basis his whole theorem utterly ceases to prove its point.

?- LP = not(true(LP)).
LP = not(true(LP)).

?- unify_with_occurs_check(LP, not(true(LP))).
false.
WHEN, and IF, you EVER PROVIDE ANY so-called 'real life' examples here, then I WILL SHOW and PROVE HOW and WHY what you want to design and achieve here will NEVER WORK.
PeteOlcott wrote: Sun May 28, 2023 2:14 pm https://en.wikipedia.org/wiki/Tarski%27 ... neral_form
In my system (having the same architecture as the Prolog inference model) the truth of expressions of language is derived from Facts and/or deduced from Facts based on Rules. The Liar Paradox (in this system) is simply untrue. Such a system cannot be incomplete in the Gödel sense because Unprovable(L,x) simply means Untrue(L,x).
Although 'godel's theorem', the 'liars paradox', and 'tarski's undefinability' are all False, Wrong, Inaccurate, Incorrect, and/or incomplete in and of themselves, it is NOT 'these things' that is STOPPING you from achieving what you want here.

What IS ACTUALLY STOPPING you is the Fact that 'it' IS an IMPOSSIBLE goal.

you are just SAYING and USING 'these things' to 'TRY TO' BLAME for NIT YET achieving what you want here.
It is not impossible for a Chatbot to have sufficiently complete and correct model of the world to very effectively argue against claims of election fraud in the 2020 presidential election.

The simple fact that there is no evidence of election fraud sufficient to change the results of the 2020 election is an excellent basis to begin with.
LOL

you say and claim here that some so-called 'chatbot' could complete some model, and then 'effectively argue', for that model above here. But, let us NOT forget that even 'you', a human being', just SAID there there was not enough sufficient evidence to change the results.

So, 'you', "yourself', have, ALREADY, JUST DONE what you say is NOT impossible for a chatbot to be able to do.

Besides this OBVIOUS Fact NO computer could make ANY such CLAIM as there is NO evidence existing, for some 'thing', if 'you', human beings, have NOT YET DECIDED, we are STILL AT, WHEN you discover and/or work out what 'it' is EXACTLY what 'truth', itself, is solely depended upon, then we can MOVE ALONG and PROGRESS here.
Age
Posts: 20043
Joined: Sun Aug 05, 2018 8:17 am

Re: Why is The Gettier problem still considered an open issue?

Post by Age »

PeteOlcott wrote: Fri Jun 02, 2023 8:43 pm
Skepdick wrote: Fri Jun 02, 2023 8:16 pm
PeteOlcott wrote: Fri Jun 02, 2023 8:14 pm In Prolog every recursive definition that it detects is pathological.
Then don't use Prolog.

This is a perfectly valid (and meaningful) definition in OCaml.

Code: Select all

let rec seq acc = function
    | 0 -> acc
    | n -> seq (acc+1) (n-1)
;;
val seq : int -> int -> int = <fun>    
It evaluates to a function which can be applied to arguments

Code: Select all

utop # let x = seq 5;;
val x : int -> int = <fun>
utop # x 0;;
- : int = 5
That is not encoding the Liar Paradox.
My purpose in showing that the actual Liar Paradox is simply a semantically incorrect expression of language is to overturn the Tarski Undefinability theorem that utterly depends on the Liar Paradox being self-contradictory.
This has ALREADY BEEN DONE. So, WHY do you want to KEEP DOING what HAS ALREADY BEEN DONE.
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Why is The Gettier problem still considered an open issue?

Post by PeteOlcott »

Age wrote: Sat Jun 03, 2023 3:19 am
PeteOlcott wrote: Fri Jun 02, 2023 8:43 pm
Skepdick wrote: Fri Jun 02, 2023 8:16 pm
Then don't use Prolog.

This is a perfectly valid (and meaningful) definition in OCaml.

Code: Select all

let rec seq acc = function
    | 0 -> acc
    | n -> seq (acc+1) (n-1)
;;
val seq : int -> int -> int = <fun>    
It evaluates to a function which can be applied to arguments

Code: Select all

utop # let x = seq 5;;
val x : int -> int = <fun>
utop # x 0;;
- : int = 5
That is not encoding the Liar Paradox.
My purpose in showing that the actual Liar Paradox is simply a semantically incorrect expression of language is to overturn the Tarski Undefinability theorem that utterly depends on the Liar Paradox being self-contradictory.
This has ALREADY BEEN DONE. So, WHY do you want to KEEP DOING what HAS ALREADY BEEN DONE.
?- LP = not(true(LP)).
LP = not(true(LP)).

?- unify_with_occurs_check(LP, not(true(LP))).
false.

Copyright 2021 PL Olcott

The above has not been done until I did it.
It conclusively proves that the Liar Paradox is semantic gibberish.
Once we understand that the Liar Paradox is semantic gibberish that can
be detected and rejected it can no longer be used as the basis of the
Tarski Undefinability theorem.
https://en.wikipedia.org/wiki/Tarski%27 ... neral_form
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Why is The Gettier problem still considered an open issue?

Post by PeteOlcott »

Age wrote: Sat Jun 03, 2023 3:13 am
PeteOlcott wrote: Mon May 29, 2023 8:16 pm
Age wrote: Mon May 29, 2023 8:11 am

But Who is going to program WHICH WAY the chat it is going to LOOK AT and VIEW 'things'?

you keep appearing to completely MISS or MISUNDERSTAND the ACTUAL ISSUE here


False.


WHEN, and IF, you EVER PROVIDE ANY so-called 'real life' examples here, then I WILL SHOW and PROVE HOW and WHY what you want to design and achieve here will NEVER WORK.


Although 'godel's theorem', the 'liars paradox', and 'tarski's undefinability' are all False, Wrong, Inaccurate, Incorrect, and/or incomplete in and of themselves, it is NOT 'these things' that is STOPPING you from achieving what you want here.

What IS ACTUALLY STOPPING you is the Fact that 'it' IS an IMPOSSIBLE goal.

you are just SAYING and USING 'these things' to 'TRY TO' BLAME for NIT YET achieving what you want here.
It is not impossible for a Chatbot to have sufficiently complete and correct model of the world to very effectively argue against claims of election fraud in the 2020 presidential election.

The simple fact that there is no evidence of election fraud sufficient to change the results of the 2020 election is an excellent basis to begin with.
LOL

you say and claim here that some so-called 'chatbot' could complete some model, and then 'effectively argue', for that model above here. But, let us NOT forget that even 'you', a human being', just SAID there there was not enough sufficient evidence to change the results.

So, 'you', "yourself', have, ALREADY, JUST DONE what you say is NOT impossible for a chatbot to be able to do.

Besides this OBVIOUS Fact NO computer could make ANY such CLAIM as there is NO evidence existing, for some 'thing', if 'you', human beings, have NOT YET DECIDED, we are STILL AT, WHEN you discover and/or work out what 'it' is EXACTLY what 'truth', itself, is solely depended upon, then we can MOVE ALONG and PROGRESS here.
A Chatbot and a human with a sufficient model of the world will both
know that there is no evidence of election fraud that would change
the 2020 election. This same Chatbot or human would also know that
there is much evidence that claims of election fraud are deliberate lies.
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Why is The Gettier problem still considered an open issue?

Post by PeteOlcott »

Age wrote: Sat Jun 03, 2023 2:46 am
PeteOlcott wrote: Mon May 29, 2023 1:26 pm
Skepdick wrote: Mon May 29, 2023 1:22 pm
And who does the stipulation?
ChatGPT was able to figure out that {baby kittens} are not any type of {ten story office building}
ONLY and SOLELY BECAUSE that information PREXISTED BEFORE that PROGRAMMED 'chatgpt' went and did what 'it' was PROGRAMMED TO DO. Which is to just find ALREADY EXISTING information. 'chatgpt' is NOT programmed to just 'MAKE UP' whatever 'it' DECIDES nor WANTS 'to make up'.

None-the-less Chatbot does fabricate fake facts and the reason for this is
currently not sufficiently understood. Some corrective action has been taken.
The goalposts of deriving a good conclusion have been moved. Now the rewards
will be at each inference step.

My recommendation is to encode a True(L, X) predicate so that the
system has some way to distinguish verified facts from pure fiction.
PeteOlcott
Posts: 1514
Joined: Mon Jul 25, 2016 6:55 pm

Re: Why is The Gettier problem still considered an open issue?

Post by PeteOlcott »

Age wrote: Sat Jun 03, 2023 2:41 am
PeteOlcott wrote: Mon May 29, 2023 1:18 pm
Age wrote: Mon May 29, 2023 8:16 am

But WHO IS the ONE who gets to STIPULATE what IS TRUE FROM what IS NOT TRUE?

For example, can ANY one 'put their hand up' for that job?
The computer does it itself.
you appear to KEEP MISSING the Fact that a computer can ONLY 'decide' based upon the ACTUAL person, who 'programmed' ' the computer'. A computer does NOT make decisions itself. A computer just 'puts out' based solely upon what has been 'put in', again by 'you', human beings. So, what this means is if there is absolutely ANY bias within the human being, programming the computer, then 'this bias' could all to easily and simply be transferred into 'the computer' as well.

The 'biases', by the way, which ALL of 'you', adult human beings, HAVE can be very easily NOTICED and SEEN.
"you appear to KEEP MISSING the Fact that a computer can ONLY 'decide'
based upon the ACTUAL person, who 'programmed' ' the computer'."

With deep learning large language models this is no longer true. Within
these models the machines are mostly self-taught.
Skepdick
Posts: 14347
Joined: Fri Jun 14, 2019 11:16 am

Re: Why is The Gettier problem still considered an open issue?

Post by Skepdick »

PeteOlcott wrote: Sun Jun 04, 2023 2:50 pm With deep learning large language models this is no longer true. Within
these models the machines are mostly self-taught.
What does that have to do with anything?

You are using Prolog, not machine learning.
Age
Posts: 20043
Joined: Sun Aug 05, 2018 8:17 am

Re: Why is The Gettier problem still considered an open issue?

Post by Age »

PeteOlcott wrote: Sun Jun 04, 2023 2:39 pm
Age wrote: Sat Jun 03, 2023 3:19 am
PeteOlcott wrote: Fri Jun 02, 2023 8:43 pm

That is not encoding the Liar Paradox.
My purpose in showing that the actual Liar Paradox is simply a semantically incorrect expression of language is to overturn the Tarski Undefinability theorem that utterly depends on the Liar Paradox being self-contradictory.
This has ALREADY BEEN DONE. So, WHY do you want to KEEP DOING what HAS ALREADY BEEN DONE.
?- LP = not(true(LP)).
LP = not(true(LP)).

?- unify_with_occurs_check(LP, not(true(LP))).
false.

Copyright 2021 PL Olcott

The above has not been done until I did it.
It conclusively proves that the Liar Paradox is semantic gibberish.
Once we understand that the Liar Paradox is semantic gibberish that can
be detected and rejected it can no longer be used as the basis of the
Tarski Undefinability theorem.
https://en.wikipedia.org/wiki/Tarski%27 ... neral_form
ALREADY DONE, so let us MOVE ON.
Age
Posts: 20043
Joined: Sun Aug 05, 2018 8:17 am

Re: Why is The Gettier problem still considered an open issue?

Post by Age »

PeteOlcott wrote: Sun Jun 04, 2023 2:43 pm
Age wrote: Sat Jun 03, 2023 3:13 am
PeteOlcott wrote: Mon May 29, 2023 8:16 pm

It is not impossible for a Chatbot to have sufficiently complete and correct model of the world to very effectively argue against claims of election fraud in the 2020 presidential election.

The simple fact that there is no evidence of election fraud sufficient to change the results of the 2020 election is an excellent basis to begin with.
LOL

you say and claim here that some so-called 'chatbot' could complete some model, and then 'effectively argue', for that model above here. But, let us NOT forget that even 'you', a human being', just SAID there there was not enough sufficient evidence to change the results.

So, 'you', "yourself', have, ALREADY, JUST DONE what you say is NOT impossible for a chatbot to be able to do.

Besides this OBVIOUS Fact NO computer could make ANY such CLAIM as there is NO evidence existing, for some 'thing', if 'you', human beings, have NOT YET DECIDED, we are STILL AT, WHEN you discover and/or work out what 'it' is EXACTLY what 'truth', itself, is solely depended upon, then we can MOVE ALONG and PROGRESS here.
A Chatbot and a human with a sufficient model of the world will both
know that there is no evidence of election fraud that would change
the 2020 election. This same Chatbot or human would also know that
there is much evidence that claims of election fraud are deliberate lies.
Who REALLY cares about some Truly INSIGNIFICANT TINY election in some Truly INSIGNIFICANT country, in some Truly INSIGNIFICANT period?
Age
Posts: 20043
Joined: Sun Aug 05, 2018 8:17 am

Re: Why is The Gettier problem still considered an open issue?

Post by Age »

PeteOlcott wrote: Sun Jun 04, 2023 2:48 pm
Age wrote: Sat Jun 03, 2023 2:46 am
PeteOlcott wrote: Mon May 29, 2023 1:26 pm

ChatGPT was able to figure out that {baby kittens} are not any type of {ten story office building}
ONLY and SOLELY BECAUSE that information PREXISTED BEFORE that PROGRAMMED 'chatgpt' went and did what 'it' was PROGRAMMED TO DO. Which is to just find ALREADY EXISTING information. 'chatgpt' is NOT programmed to just 'MAKE UP' whatever 'it' DECIDES nor WANTS 'to make up'.

None-the-less Chatbot does fabricate fake facts and the reason for this is
currently not sufficiently understood. Some corrective action has been taken.
The goalposts of deriving a good conclusion have been moved. Now the rewards
will be at each inference step.

My recommendation is to encode a True(L, X) predicate so that the
system has some way to distinguish verified facts from pure fiction.
As I have ALREADY PARTLY EXPLAINED what you want here can NOT happen.

So, let us, AGAIN, move along here.
Age
Posts: 20043
Joined: Sun Aug 05, 2018 8:17 am

Re: Why is The Gettier problem still considered an open issue?

Post by Age »

PeteOlcott wrote: Sun Jun 04, 2023 2:50 pm
Age wrote: Sat Jun 03, 2023 2:41 am
PeteOlcott wrote: Mon May 29, 2023 1:18 pm

The computer does it itself.
you appear to KEEP MISSING the Fact that a computer can ONLY 'decide' based upon the ACTUAL person, who 'programmed' ' the computer'. A computer does NOT make decisions itself. A computer just 'puts out' based solely upon what has been 'put in', again by 'you', human beings. So, what this means is if there is absolutely ANY bias within the human being, programming the computer, then 'this bias' could all to easily and simply be transferred into 'the computer' as well.

The 'biases', by the way, which ALL of 'you', adult human beings, HAVE can be very easily NOTICED and SEEN.
"you appear to KEEP MISSING the Fact that a computer can ONLY 'decide'
based upon the ACTUAL person, who 'programmed' ' the computer'."

With deep learning large language models this is no longer true. Within
these models the machines are mostly self-taught.
you KEEP MISSING the POINT.
Post Reply