WotC Would you buy WotC products produced or enhanced with AI?

[briggart]

The assessment of the doubt may involve probability theory--but that isn't science, it's mathematics, which is built on logic, not on empiricism. Empiricism has so thoroughly internalized mathematics that some forget that that isn't where probability theory came from. It's not experiment which proves statistics are useful--it is statistics which prove experiments useful. Where does statistics derive its truth from? You can't use statistical modeling to explain why statistical modeling explains things!

The role of math and logic in science is not dissimilar to that of grammar in ordinary communication.

"Outside my house, elephants are flying in a sky filled with flowers" it's a perfectly valid sentence from a grammar point of view, but that by itself is not enough to say if it is also an accurate description of reality. That requires somebody going outside my house, and looking at the sky.

A more scientific example: the math of Newtonian gravity works perfectly fine with two mass signs. The conclusion that there is no negative mass is purely an empirical one.

Math and statistics can tell you how to properly combine your observations of the world to get to the least misleading conclusions, but cannot tell you the results of those observations in the first place.

 

[EzekielRaiden]

The role of math and logic in science is not dissimilar to that of grammar in ordinary communication.

"Outside my house, elephants are flying in a sky filled with flowers" it's a perfectly valid sentence from a grammar point of view, but that by itself is not enough to say if it is also an accurate description of reality. That requires somebody going outside my house, and looking at the sky.

A more scientific example: the math of Newtonian gravity works perfectly fine with two mass signs. The conclusion that there is no negative mass is purely an empirical one.

Math and statistics can tell you how to properly combine your observations of the world to get to the least misleading conclusions, but cannot tell you the results of those observations in the first place.

And where is the empirical justification of using mathematics for this purpose?

If there is no such empiricial justification, then mathematics is generating truths without reference to empirical method, which contradicts the claim that empirical methods are the sole valid route to truth. If there is such, then mathematics isn't actually built on logic...which is what it explicitly claims to be, so mathematics is already false by its own lights.

I don't think this is a particularly solvable conundrum unless we accept that mathematics can lead us to truths that are independent of any form of empiricism. Or, I guess, denying that mathematics has any validity at all, but I doubt you want to go that route any more than I do.

 

[briggart]

And where is the empirical justification of using mathematics for this purpose?

If there is no such empiricial justification, then mathematics is generating truths without reference to empirical method, which contradicts the claim that empirical methods are the sole valid route to truth. If there is such, then mathematics isn't actually built on logic...which is what it explicitly claims to be, so mathematics is already false by its own lights.

I don't think this is a particularly solvable conundrum unless we accept that mathematics can lead us to truths that are independent of any form of empiricism. Or, I guess, denying that mathematics has any validity at all, but I doubt you want to go that route any more than I do.

Science is the empirical justification for using math for this purpose.

But I feel we may be using "true" in a different way. If your requirement for a statement to be true is only internal self consistency, I agree that math does not require any empirical evidence.

But if we are asking about the actual ownership of Plato's works or about reaching a verdict in a trial, your examples which started our exchange, we are not satisfied by our conclusions being self consistent, we also want that they match our observations. Which by definition requires empirical methods. It's not an either/or situation, we need both.

Another way to put it is that math/logic can at best* define the set of all statements about our world than can be true. Empirical observations are required the identifying the subsets of those statements that are actually true.

*Math, or at the very least some of its fields, is axiomatic: e.g. Euclid's postulates. We need to assume some fundamental facts to be true to properly build math, some of which are based on empirical evidence. And this is assuming math can be complete.

 

[Lanefan]

The assessment of the doubt may involve probability theory--but that isn't science, it's mathematics, which is built on logic, not on empiricism. Empiricism has so thoroughly internalized mathematics that some forget that that isn't where probability theory came from. It's not experiment which proves statistics are useful--it is statistics which prove experiments useful. Where does statistics derive its truth from? You can't use statistical modeling to explain why statistical modeling explains things!

You can, however, use stastical modeling to explain why all the other modeling methods don't explain things; and eventually leave statistical modeling as the last method standing through process of elimination. In other words, it explains itself by explaining away all the other options.

 

[Cergorach]

Where is the 'maybe' option? It's either Yes, No, or probably No. This is called 'guiding' polls.

For me it depends. It's not as if I can't make an adventure or a setting, but I still bought oodles of WotC D&D books. And some people claim, if it's done with AI, you can do it yourself! Nothing has changed in that regard when people wrote and illustrated D&D books, we could do that ourselves as well. The issue being that writing a 100-300 page book/adventure takes a LOT of time, not to mention the illustrations and maps (even with great mapping software and premade assets).

Doing the same, but with AI/LLM is faster then doing it by hand, but still takes oodles of time. And if you want to do it well, chances are that it'll cost money as well. So, if WotC makes something with AI/LLM it depends on how well it's done, how much time it will save me, and how much it will cost. It's that simple.

 

[EzekielRaiden]

You can, however, use stastical modeling to explain why all the other modeling methods don't explain things; and eventually leave statistical modeling as the last method standing through process of elimination. In other words, it explains itself by explaining away all the other options.

That...doesn't follow. Like you've made it circular for one thing (statistical modeling can evaluate its own ability to evaluate truth?), and for another, *until you've got some metric for saying that a thing definitely does or definitely doesn't, you can't use statistical modeling to eliminate its competitors.

An assertion can work for a trillion, trillion values and still be wrong on the trillion-trillion-and-first. The fact that the sun rose yesterday and the day before and the day before that cannot be used to say that therefore statistics always works. That's...not something that follows, not even by statistics' own rules!

In more specific terms: statistical modeling is probabilistic. It cannot tell you *for sure that anything, even itself, is correct/true. Hence, it is both circular and self-defeating to use it to evaluate itself, and until you have it evaluated, its pronouncements aren't useful yet for determining truth!

 

[Sulicius]

Where is the 'maybe' option? It's either Yes, No, or probably No. This is called 'guiding' polls.

For me it depends. It's not as if I can't make an adventure or a setting, but I still bought oodles of WotC D&D books. And some people claim, if it's done with AI, you can do it yourself! Nothing has changed in that regard when people wrote and illustrated D&D books, we could do that ourselves as well. The issue being that writing a 100-300 page book/adventure takes a LOT of time, not to mention the illustrations and maps (even with great mapping software and premade assets).

Doing the same, but with AI/LLM is faster then doing it by hand, but still takes oodles of time. And if you want to do it well, chances are that it'll cost money as well. So, if WotC makes something with AI/LLM it depends on how well it's done, how much time it will save me, and how much it will cost. It's that simple.

Yes, but only if (clause) was supposed to take into account for the maybe’s.

I think your answer falls squarely in that option.

 

[Lanefan]

That...doesn't follow. Like you've made it circular for one thing (statistical modeling can evaluate its own ability to evaluate truth?), and for another, *until you've got some metric for saying that a thing definitely does or definitely doesn't, you can't use statistical modeling to eliminate its competitors.

An assertion can work for a trillion, trillion values and still be wrong on the trillion-trillion-and-first. The fact that the sun rose yesterday and the day before and the day before that cannot be used to say that therefore statistics always works. That's...not something that follows, not even by statistics' own rules!

If you're trying to suggest that we can't define as truth the statistically-ironclad fact that the sun will rise tomorrow, I don't think many will subscribe to your paper.

In more specific terms: statistical modeling is probabilistic. It cannot tell you *for sure that anything, even itself, is correct/true. Hence, it is both circular and self-defeating to use it to evaluate itself, and until you have it evaluated, its pronouncements aren't useful yet for determining truth!

Can it categorically* define something as false? If yes, it can in fact go the Sherlock Holmes route by eliminating all the other possibilities until only one remains.

* - I initially typoed that as "datagorically", which is now a word I have to find a good use for.

 

[mamba]

Science has nothing to verify either!

yes it does, you have a hypothesis, and under that hypothesis something is true, then you set out to check if that thing is in fact the case

It does not mean your explanation is entirely accurate, but if that thing is not the case, then your hypothesis is false. Science checks its conclusions / predictions all the time

You have staked out the position that science is the one and only valid path to truth.

the only reliable path to it

How do you aim to prove this claim?

it is widely accepted, your opinions notwithstanding. First you tell me law and textual analysis find truths and now that science cannot verify predictions… that does not give me confidence in your opinion

He was completely wrong though. He said his equations described existence. They don't. They are only an acceptable approximation, and that only in a narrow slice of reality.

the narrow slice that covers everything happening on Earth, seems plenty useful to me, and calling that completely wrong feels like quite the exaggeration too

 

[Cergorach]

If you're trying to suggest that we can't define as truth the statistically-ironclad fact that the sun will rise tomorrow, I don't think many will subscribe to your paper.

Well, almost 175k people won't see sun rise tomorrow. The sun is still there, the earth is still there, but that won't always be the case. We know that won't always be the case. There could be some unexpected events that could make that happen a heck of a lot sooner though. But those kinds of things are only interesting from a personal perspective... So the question is from which perspective will the sun 'rise' tomorrow?