SatanLucy

Finding objective truth simply means starting from logical law as a premise, and as long as someone accepts that logic even exists and isnt too stupid, they must accept premise 1 because premise 1 is merely a basic logical law.


Good example of premise 1 is:


P1. A is true or Not A is true


there is no way to disagree with this premise, without denying that logic even exists.


So as long as someone accepts that logical laws exist, any first premise made in form of logical law itself must be accepted as true.


Another example of objective truth is definitional conditional premise:


P1. If definition of A is B, then A is B


this premise remains true by law of identity.


third, weaker form of objective truth is observational premise.


P1. We see that A happens with B


But this one can be denied by questioning validity of observations



But objective truth isnt gained by debating, or polls, or "experts". these entirely depend on who decides what truth is. A perfectly reasonable argument can be rejected, while unreasonable one may be accepted, depending on who the "judge of truth is".


the main problem in determining truth is that such right cannot be exclusive to anyone. In same way reciprocity works with morality, it likewise works with truth determining. If you say "We decide what truth is", I can simply say "No, I do.".

SatanLucy
If someone is anti-social and behaves like a twat


I dont behave like a twat. I am a nice person. Not because I think being nice is always good by default, but because it benefits my self-interest and improves my position in society through reciprocity principle which holds valid in majority or at least a great amount of human relationships, much greater than being not nice.

SatanLucy
Only a therapist or a psychiatrist can diagnose you with anti-social personality disorder and give you insight what's going on in your head.


thats only true if I tell therapist all my thoughts, which is a bad idea for me.

SatanLucy
If your product causes harm, you have liability meaning you can be sued, doesn't mean you'll lose the case.


Everyone can be sued for anything. that is not definition of liability.


Liability is usually defined as legal obligation to pay or provide.


But justification for AI company being liable or having liability to provide anything due to harm AI causes doesnt exist.


Because justification for AI advancement outweighs justification for punishing AI companies for this.


So your "absolute claim", which then degraded to "not true in many cases", likewise ceases in case of AI, because advancing AI is more important than placing endless restrictions to sterilize its growth.


So I am afraid the agreement which you think exists between us exists only in your head.

SatanLucy

No one knows what happens after death. tho a lot of people pretend they know with pure certainty something which they have no way of observing or logically deducing.

SatanLucy

Damn

SatanLucy · 6 months ago

Realistically, the only way world ever improves is if I somehow become supreme unconditional leader. Until then, it remains poop.

SatanLucy

Damn

SatanLucy · 6 months ago

I dont care for politics. People who care for politics are in for a disappointment. Elections are controlled and binary. Its sort of like someone offering you two different types of poop. And sure, one can say "this poop is better than the other poop". Still a poop.

SatanLucy
And no one has argued anything having to do with "all cases"


You:

"When you create a product for public consumption you have a responsibility to ensure that product comes with the proper safeguards in place to ensure it is not utilized in a way that causes harm."


This was an absolute claim, not "in many cases".


And if it wasn't an absolute claim, then you can't tell if it applies to AI or not.

SatanLucy

the biggest issue for me isnt how to make chatbot coherent. that is easy, as shown in that coherent bot version.


the problem is, how to make chatbot form entirely new sentences while remaining coherent.


Repeating from training data and being coherent is easy. Using training data to make something entirely new not found in training data while keeping coherence? thats where real trouble starts for a simple chatbot. Its usually a dichotomy between coherence plus repetition or creativity plus incoherence.

SatanLucy
Your AI output is mostly gibberish.


Because it forms new sentences not found in training data, and of course not all of those sentences make sense.


Sure, I could make it merely repeat training data, and that would be 100% coherent, but then no creativity happens.


this is example of "coherent" bot:




So yeah, this one is "coherent", but problem in it is that it merely repeats from data. Produces nothing new. So it can never produce a single sentence not found in its training data.

SatanLucy

this is first model:




this is second model with smaller training data:




this is 3rd model, with huge training data and huge pattern groups:




3rd model is essentially huge. Its brain txt file is almost 500 megabytes.


Whats interesting is that it produced knowledge which isnt found anywhere in its training data. It combined "invisible hand self interest theory" to "debating", and as result, produced argument that appealing to self interest is an efficient debating strategy.


this AI doesnt use weights, and no it doesnt use markov chain either.


I used to play with markov chain, but it can never be good because it either produces incoherence or merely copies training data.


this AI tho, produces coherence without copying training data itself. It uses patterns from training data to form new sentences not found in training data itself.


And training time is extremely fast, it takes program about few seconds to make its fully ready "brain.txt" file it uses for chatting. It makes "brain.txt" file from training data.


this is the most advanced AI I have ever built. And the one which works surprisingly well, with very simple coding.



SatanLucy

The main advantage of "weights based" AI models is their reasoning ability, which in some cases fails even at top models.


AI which isn't based on weights, but by pure next word prediction, can actually become creative too and write things which aren't contained in it's training data.


Because predicting next word doesn't actually equal copying whole sentences. In fact, if set to maximum randomness of next word other than following after current word somewhere in text, it will almost guaranteed produce no sentences found in its training text at all. Almost nothing it produces would match training data.


Weights based AI are extremely inefficient at a lot of what they do right now.


On one hand, they are given millions of texts to learn from.


On the other hand, they are not always learning effectively and sometimes even contradict training data.


And also, they are expensive. Their training takes long, they consume bunch of resources, and money.


Local AI are cheaper to run, but here lies the problem. Local AI are already trained. So you are getting an AI someone else trained. Unless you code it yourself, but training it is difficult because to train AI takes a lot of text.


So I came up with a very interesting solution.


An AI which is capable of not only learning from any book you give it, but also forming new sentences which aren't even found in that book.


I used to code chatbot which merely retrieves text from book.


But I realized I can actually code AI which isn't a mere copy pasting machine.


AI based on predicting next word, while keeping certain level of randomness, actually is capable of generating entirely new sentences because it produces word by word, and is able to combine parts or word groups of different sentences into one new sentence without creating a mess.


Should finish coding it tomorrow.


This is perhaps the most advanced AI I ever coded, because logic behind it is crazy. It allows both coherence and creativity, so it can make perfectly coherent new sentence not found anywhere in its training data, and at same time remain coherent thanks to context window.


I think such AI is better, because it's extremely light weight. It learns patterns entirely from groups of words, so it can combine groups into new sentences.


Because it's light weight, its much faster than any local AI, and at same time can be fed any training data in text form, any book or whatever.


While it's not as smart as AI which uses weights, it makes up for that by being more precise. It doesn't "hallucinate" in any sense of that word unless person manually increases randomness. New sentences it makes still come from word groups in it's training data. It doesnt "make something out of nothing".


I will see if I can finish this project.

SatanLucy
Which translates into "in many cases yes". So we're in agreement.


Your original claim didn't include "many cases" but was an absolute claim. And no, we are not in any sort of agreement on the issue. Do you see "I agree with you" written anywhere in my previous comment?