SatanLucy
At one point, he was minutes away from a permanent ban for advocating pedophilia, name change and a new website saved his sorry ass.


Not really. I was never in danger of permanent ban. In fact, my ban was self-requested because I wanted to change my username and needed to make new account to do it. That's why I requested Vader to ban me, and he did.




Small wonder the pedophile is the number one champion of free speech, to him Free Speech is a religion LOL


If you hate free speech, that's okay. It means you hate US constitution.

SatanLucy
Free speech is sacred


Well, duh. I dont really see how you build a site where people express themselves, while banning freedom of expression.


I also dont see how you maximize expression while reducing freedom of expression for arbitrary reasons.


Maybe sometimes, reducing some forms of expression can yield higher expression overall, but I really dont see what logic person uses to calculate that.


Also, I am all for reciprocity principle. Remy Brown didnt take away anyone's free speech, so trying to take his free speech away is not only grand injustice, but clearly not eye for eye which is basis of moral justice.

SatanLucy

He made free speech win on this site. As I said before, trial was about Remy Brown, but it wasnt trial only about Remy Brown. It was trial about anyone out there who holds views masses consider offensive, because outcome of trial sets a principle which applies further than one specific user.

SatanLucy

That's expensive, but you can't run windows programs on it. It's reason why I would never get apple computer. So many games don't work on it.

SatanLucy

I guess history will prove if you are right. Currently, I am busy testing out new laptop so I have less time to make these arguments.

SatanLucy

Its gonna take me some time to download vampire masquerade bloodlines 2. Game is 20 GB from what I see. However, its very demanding, so i wanna test if this laptop can play it. RAM isnt the issue. dual channel RAM of 16 GB runs anything. Its the graphic card which could be insufficient. 4 GB VRAM, but this game demands 6. My integrated graphics has 8 GB VRAM, but thats borrowing from RAM.

SatanLucy

Typing this on my new laptop. Its very good. Bigger screen too. Also runs skyrim special edition on 60 fps on max settings. Good enough for 140$.

SatanLucy

Topic 1

SatanLucy · 2 months ago

I am all for anarchy, but its still mods who decide cuz well, they own site or exercise will of owner.

SatanLucy
This is probably best explained by the paperclip maximizer thought experiment. If an ASI has an arbitrary goal (like maximizing the number of paperclips), it can do a better job of this if humans cannot shut it off. So the actions that work best toward an arbitrary terminal goal include either permanently depowering humans or perhaps killing them all.


I mean, there are a couple of issues with this assumption.


Your argument is basically:


AI has a goal = it can do that goal better if it exterminates humans


But this depends on goal entirely. AI can also have goal to protect humans.


Another problem with the argument is that current AI models don't have "goals".


It may seem like a goal driven behavior, but it's not goal at all.


What AI currently does is, well, AI is a calculator.


It predicts next token based on context.


What happens when you put your input in AI?


AI converts your input to numbers (vectors), multiplies them by many weights in a non linear way (because linear multiplication is useless).


It then produces next token.


And then that token and your input both become context, converted into numbers, multiplied again, to predict next token.


It may seem sentient. It may seem smart. It may seem like it has a goal.


Reality is that it's a calculator for probability of next token.


Maybe future AI won't be like this, but current AI is literally just calculator which calculates probability of next token.


Weights which you talk about are just numbers it multiplies tokens by, and those get changed during training for better prediction of next token.

SatanLucy
I think even you as a critic of induction wouldn't put it on the level of dice divination...


In any case where you don't have exact probability, that's what it comes down to.


If something is 60% true in observed cases, and most cases are not observed, there is no telling what probability would be if most cases were observed.


And these situations happen often in social science. I mean, the polls were at some point suggesting Trump will lose elections. Problem lies in the potential error.

SatanLucy

Soon. There is still time for you to cast vote on that debate if you want. I think it's safe to say Con won at this point. But more votes is better because it creates greater majority and affects future mod decisions more.

SatanLucy
95% of accidents occur in or within 5 miles of home


5 miles away from home is a pretty dangerous distance. Why would someone travel so far from home?

SatanLucy

Actually, the safest option is to not go outside.


I assume with all those shootings happening, a lot of Americans are considering bedrotting lifestyle as best alternative.


I don't really see point of going outside, from the perspective of safety. Reality is, you have reasonably high chance to be involved in car accident by just being outside a lot.


As for women in USA, given that USA has over 1 million sexual abuse cases per year, if I were a woman in USA, I would not go outside.

SatanLucy

Also, I have to make clear difference between deductions based on observations and induction based on observations.


Deduction based on observation is:


"this applies in all areas we observed, so it applies to this observed area as well"


this logic is always true, because specific observed area is part of "all observed areas". Conclusion must follow from premise.


Induction goes beyond observed area and says:


"this applies in all areas we observed, so it applies to all non-observed areas as well"


this is a leap of faith. conclusion doesnt follow from premise.

SatanLucy
The goal being to improve accuracy, right? You must recognize deduction can yield absolute certainty in a faulty conclusion. All it takes is a bad premise.


If we are talking of actual deduction, then the foundational premise must be generally true claim.


It's true people can use deduction to arrive at incorrect conclusion, but that's only if their premises are incorrect and also not themselves deductions. It's actually possible to use deductive reasoning in flawed form, most famous one being "if, then" form, which is actually more often misused than used accurately.


The basic idea around deduction is to take part from the whole.


The classic example is:


"All A are B, so some A are B".


This is logic moving from general to specific.


And as far as I know, this logic cannot produce false conclusion unless "All A are B" is false.


How one comes to premise "All A are B" is a different thing. It can happen by definition, logical law, exhausting options, or observation(induction).


But when it happens by induction, you are again looking at weaker argument unless it's a strong constantly repeated observation.


Yes, the goal is to improve accuracy, and we can say for sure proper deduction is true as long as premises are true. Induction on the other hand can be false even if premises are true. Induction presented in form of deductive logic suffers from unproved premise from start.



Induction to spot the patterns and make the premise, and deduction to predict what happens if its true


You are correct that induction is often needed to make a premise. Not always. Sometimes starting premise can merely be logical law or exhausting options and exhausting explanations to see which explanation remains both internally consistent and not contradicting some true theory elsewhere, and which explanation is absurd.


My issue isn't with strong induction. My issue is more with relying too heavily on induction to the point where it's used even in cases where it has decently high chance to be wrong.


I don't deny that laws of physics which were generalized by induction are true. Maybe they are not true everywhere in universe, because they were derived by induction, but the chance of them being true is more likely than not in some area.


My issue is more with using induction beyond it's probable limits. We also know there are cases where induction is terribly uncertain. We know that AI uses induction to produce hallucinations.


I can agree that induction can be useful. My point of disagreement would be that induction should really be a last resort, especially in cases of extrapolation, which is least reliable formal form of reasoning there is.


Science which prefers induction over all other forms of reasoning is basically guaranteed to make bunch of false conclusions. And we have seen that with pretty much any science which relied on induction, be it psychology, sociology, pharma drug testing, social studies...ect.


We see flaw of these sciences because they are using form of reasoning which is considered nonsense in formal logic.


This is not to be confused with strong induction. Strong induction is basically something which repeats as exactly true each time when observed, even with expansion of observation.


Soft induction is area where results vary, so claim is sometimes true, sometimes not true. A lot of social science suffers from this issue.


Strong induction does have its own problems, such as assumption that what we observed so far follows same laws as remaining 99% of things we didn't observe.


However, at least in case of strong induction, we assume it's 100% true in areas we observed so far, so we can apply it there, and I don't think it's likely that laws of physics will stop working tomorrow.


However, even strong induction, when applied to non-observed areas, becomes flawed. I mean, black holes are still partly a mystery despite all strong induction we had so far.

SatanLucy

Even AI agrees with me:


"Yes, extrapolation is generally considered the least reliable method for making predictions because it estimates values outside the range of your actual measured data"


AI should know, given that it's main issue AI faces today. They try to solve this issue by increasing training data, and increasing parameters count to make AI smarter so it doesn't hallucinate as much due to this issue.

SatanLucy

As Savant mentioned, extrapolation based on known distribution is unreliable even for AI or computers. Interpolation is more reliable. However, inductive reasoning based on small percentage of observed world is basically extrapolation. Interpolation allows deductive reasoning, because it's usually just filling gaps in training data. But the greater the unobserved world compared to observed, greater the gap.


Extrapolation induction is about as unreliable as it gets in science. The only next step would be making predictions based on dice roll.

SatanLucy
What are you advocating? ...changing science so it doesnt allow more information to hone accuracy, or, what, getting rid of science?


What I advocate for is greater focus on deductions, observations and probabilities made by those.


I really dont think science which are based entirely on inductions have any great future, because their whole rule for logic they use is fundamentally flawed and proved false by countless examples in countless scenarios. there is a reason why sociology and psychology are called "soft science", because they rely on induction and its pretty much universal knowledge in scientific world that induction is not just uncertain, but sometimes even blatantly misleading.


there are only few areas where inductive logic is good, and those are always consistent repetition of conclusion. Because inductive reasoning doesnt use deduction, its conclusion entirely depends on observations to be proved likely true over and over in each observation.


However, in any case where observations are inconsistent or largely incomplete, conclusion made by induction resembles more of an assumption than reasoning.


I say, sure, if you have nothing else, use induction. But really, order of work should be deduction > observation > explanation exhaustion testing > induction. I guess definition deduction also have their place, and some science is based entirely on definition deduction, like math. However, definition deduction rarely applies to other forms of science, because it does require definitions based on primitives as opposed to arbitrary definitions. Math definitions are based on primitives. Even if you changed definition of "2", the primitive rule of addition would actually still be true. So definition deduction mainly deals with definitions made from primitive concepts. It descends into inconsistency if definitions arent based on some fundamental rule of logic. Which also does bring us to starting premises being laws of logic, and other than "A = A", the basic rule of logic is deduction (All A are B means "single A" is B).


As I said before, induction goes contrary to deduction, and its basic rule is that "Some A are B means All A are B". this rule is proved easily false, even in case of probabilities where "In some cases, probability is 50% = In all cases, probability is 50%" is also proved false.


In other words, induction uses a form of logic which can be true, but at same time, it can be false, so it takes a lot of observations to establish probability, and right now humans arent really in position to observe everything there is, to increase chances of induction logic being correct.

SatanLucy
These aren’t problems for science


Lack of certainty or known probability is a problem.



Absolute certainty is itself impossible to logically maintain unless you begin with the presumption that you are infallible


I mean, deductive logic is literally certainty of conclusions from premises. Math which works on deductive logic is literally certainly true. I don't really see why you even bother to respond if this will be your response.