Unfortunately, it's almost inevitable now that Al will become in the near future THE general authority. So, thinking will no longer be a practical necessity. We could even draw a logical line from human laziness to a situation where people simply plug their "personality" into a mobile AI, stick it on themselves, and allow it to do all their conversing for them. — Baden

I very much agree.

All we can do is be the change we want to see. — Baden

Okay, fair. Still, I want to say that the canons for reasoning that we have developed as a species are reliable and recognizable. What is at stake now is a particular kind of appeal to authority: appeal to LLM. Our canons include sound principles for determining when an appeal to authority is permissible and when it is not, but such principles will be challenged by the advent of AI.

TPF already has a precedent for disallowing or at least discouraging certain sources for appeals to authority, particularly sources which are deemed morally inappropriate . I would suggest that it is at least possible to establish a precedent for discouraging the "appeal to LLM" move, especially given the soft and flexible nature of TPF rules. And perhaps the current rule already does this to some extent.

(More specifically, I would say that our canons for reasoning generally require that the inferential steps used to reach a conclusion be made publicly available. A post which leverages an LLM in a way that is consistent with this principle would not be beyond the pale, given that such a post would not merely be appealing to the LLM as a blind authority. Yet a post which relies on an LLM in a way that is "blind" and inconsistent with this principle would be beyond the pale.)

I don't think ALL beliefs are IBEs:
-We have some basic, intrinsic beliefs, that aren't inferred. Example: the instinctual belief in a world external to ourselves.
We also accept some things uncritically (no one's perf ect). — Relativist

Okay good, so let's look at this quote of yours:

I've been treating "underdetermined" as any belief that is not provably true (i.e. determined=necessarily true). Under this extreme definition, nearly every belief we have is underdetermined...

Most of our beliefs are not provably true, so I have labelled them IBEs. — Relativist

In that last sentence you seem to equate "not provably true"/undetermined with "IBE." Now you are implying that something that is not provably true might not be an IBE if it is not inferred. Do you stand by your decision to label beliefs that are not provably true IBEs?

The more central question can be restated with your claim, "Under this extreme definition, nearly every belief we have is underdetermined." The "nearly" makes me think that some beliefs are not underdetermined, but I'm not sure if you really hold that.

Re: certainty- that's an attitude, and it may or may not be justified. — Relativist

In English "certitude" connotes subjectivity, whereas "certain" and "certainty" need not. When I said, "premises which are foundational and certain," I was using 'certain' in this objective sense, which is quite common. For example, you that some beliefs follow necessarily from other beliefs/facts. I might ask, "But do they really follow necessarily?" You might answer, "They certainly do." Your answer would not mean, "I have a high degree of certainty or a high degree of certitude that they do follow necessarily." It would mean, "They objectively follow necessarily."

Now if you want to try to find a different word than "certain" to describe the phenomenon in question, you can do that. I think certain is actually the correct word, given that it couples the knower and the known in the proper way needed to speak about first principles.

Justification doesn't require deductive proof. Consider your example "this entity before me is either a tree or it is not a tree." Solipsism is logically possible, so that there actually isn't something before you. We can justifiably feel certain despite the logical possibility we're wrong. — Relativist

Well the same problem crops up here. What is certain and a feeling of certainty are not the same thing, just as justification and a feeling of justification are not the same thing. One's being justified and one's deeming themselves justified are two different things.

An adherence to merely syllogistic logic might explain some of the difficulties had hereabouts. — Banno

It has many ways of dealing with many placed predicates and relations. The ancients and medievals did not lack a notion of polyadic properties. Indeed the core sign relation for language, supposition, and epistemic relations are all triadic. — Count Timothy von Icarus

Gyula Klima recently made available his contribution to the The Cambridge Companion to Medieval Logic, "Consequence." It offers a helpful remedy to the historically ignorant opinion that Medievals did not study non-syllogistic forms of logic.

Good question. We have beliefs that follow necessarily from other beliefs/facts, so they're provable in that sense. It seems inescapable that we depend on some foundational beliefs. So nothing can be proven without some sort of epistemological foundation. What are your thoughts? — Relativist

That seems right to me. In an Aristotelian sense we would speak of demonstrative arguments and non-demonstrative arguments, where demonstrative arguments have premises which are foundational and certain, whereas non-demonstrative arguments have premises which are non-foundational (and therefore also not certain). We could also apply that distinction to the inferences rather than the premises if we are not counting an inference as a premise.

So when you said:

Most of our beliefs are not provably true, so I have labelled them IBEs. — Relativist

...I guessed that this meant that some of our beliefs are provably true, and are not IBEs. If that guess is correct, then apparently you must hold that some foundations are certain. If that guess is wrong, then it would seem that you hold that all beliefs are (unprovable) IBEs. Do you disagree with any of that?

Now for Aristotle one certain foundation that we are capable of having is the principle of non-contradiction, and therefore an example of a provable conclusion would be, "This entity before me is either a tree or it is not a tree."

If some people do things for no reason then your approach works. But no one does things for no reason. No one fails deliberately. — Leontiskos

...And what's interesting here is that the foil that many desire would be highly problematic. If people really did things for no reason, then it would not be possible to convince them or anyone else that things should not be done for no reason. This is because in order to persuade someone to act differently, you supply them with reasons to change, and someone who acts for no reason is immune to such persuasion.

This is the paradox that so few seem to understand, and it applies to all forms of evil/error/sin. It is part of the mysterium iniquitatis.

No, that's just one example of Evil and not all examples. — Barkon

My point is that that example highlights the difficulty with your whole conception. If some people do things for no reason then your approach works. But no one does things for no reason. No one fails deliberately.

If you want to think about it differently, every time two people disagree over what to do, they are disagreeing over what should be valued. They both think that the other person is acting in a sub-optimal or "evil" manner. Neither one is acting for no reason or failing deliberately.

Of course, one can make a case for the existence of malice, but it requires philosophical work.

This is no different than having your friend do your homework for you. If he explains you the topic, you read the book, you understand it, you do the assignment, you're fine. If he does it for you, then you cheated, and no one likes a cheater. — Hanover

AI is result-oriented. Intellectual development, and particularly philosophical intellectual development, is process-oriented. If you just want to post the "right" answer, you are doing things wrong. — Baden

I'd say that what is inevitably going to happen (and is already beginning to happen on TPF), is that folks are going to appeal to LLMs as indisputable authorities. "You say X but my almighty LLM says ~X, therefore you are wrong." This will occur explicitly and also in various implicit ways.

Because this is an appeal to an LLM it doesn't directly contravene the rule. Nevertheless, I would argue that it is still remarkably contrary to the spirit of philosophy. It is that look-up-the-infallible-answer routine, which is quite foreign to philosophy (and is itself based on an extremely dubious epistemology).

I hope TPF will discourage this "look up the infallible LLM answer" approach, especially as it becomes more prevalent. The risk of such an approach is that humans become interpreters for AI, where they get all their ideas from AI but then rewrite the ideas in their own voice. Such a result would be tantamount to the same outcome that the current rule wishes to avoid.

(NB: The very fact that so many do not understand why a philosophy forum is intrinsically incompatible with AI-generated posts demonstrates how crucially important administrators and moderators are.)

Such as by taking a break mid sentence with no good reason, to add an insult, for all readers and who you're talking to, to decipher. It’s a bailing (like from a skateboard) with all your intention being channeled into a maleficent activity. — Barkon

I think the trick is to say what is meant by "a maleficent activity," which goes hand in hand with your idea of doing something for "no good reason." So the counterargument is as follows:

1. Evil is doing something for no good reason
2. No one does things for no good reason
3. Therefore, Evil does not exist

Or else:

4. Evil is deliberate failure
5. No one engages in deliberate failure
6. Therefore, Evil does not exist

(See also my thread, Beyond the Pale.)

Underdetermination is the theory that theories are not determined by the evidence, but rather are chosen in order to organize the evidence, and in some way are a selective pressure on which evidence is relevant to consider. — Moliere

This sounds to me a bit like post hoc rationalization, as if one is going to decide on a theory and then allow their theory to be "a selective pressure on which evidence is relevant to consider."

The difficulty here is that you seem to be redefining "theory" to be something that precedes rather than follows after evidence, and such is a very strange redefinition. For example, on this redefinition someone might say, "I have a theory...," and this statement would be indistinguishable from, "I have a prejudice..." The basic problem is that 'theory' and 'prejudice' do not mean the same thing. We distinguish between reasoning and post hoc rationalization, and yet your definition seems to have made such a distinction impossible. It seems to have made impossible a distinction between "following the evidence where it leads," and, "engaging in selection bias in favor of some a priori theory."

---

Now I'll go this far: If underdetermination, as a theory, leads us to be unable to differentiate between science and pseudo-science, and we believe there is such a thing as pseudo-science (I do), then we're in a pickle. — Moliere

I think this is one of the places where the problems become more apparent. For example, if underdetermination requires that there be multiple possible and inadjudicable theory-candidates, and nevertheless pseudoscientific theories do not belong to this set of viable candidates, then there must be some real way to separate out the wheat from the chaff. Even if one thinks this is possible they have already abandoned full-throated underdetermination in favor of an underdetermination that is nevertheless determinate vis-a-vis determining which theories are scientific and which are pseudoscience. They are doing something akin to "stance underdetermination," which is a species of a, "Underdetermined subset theory." I.e., "Within this specific subset a quasi-global underdeterminacy holds, but apart from that subset it does not hold." All of these theories struggle mightily to say how or where the specific subset ends and the complement-set begins. The task is so difficult that few such proponents even really attempt to answer that challenge.

- Okay, thanks for that. Makes sense.

Most of our beliefs are not provably true, so I have labelled them IBEs. — Relativist

So would you say that some of our beliefs are provably true?
(I would say that, but I am just verifying that you would also say such a thing.)

What I Ask of You

This position is no where near complete; and I would appreciate it, though, if people could engage with me on this position and its claims to help further or kill the ideas in it.

For those that are interested, <here's a link to the document I am writing>. — Bob Ross

This is an interesting endeavor, Bob Ross. :up:

I think the most fruitful things to pursue would be those things where you disagree with traditional Christians, in particular over whether some doctrine is accessible through natural reason (i.e. apart from revelation).

With that in mind, this seems like the most difficult thesis:

5. It provides a perfect synthesis of justice and mercy that necessitates the practical and reasonable acquisition of salvation; — Bob Ross

The difficulty here is that "salvation" is often understood as a Christian term, and in that context it is not something we can achieve on our own power. Supposing you are not using the term in that way, I would want to know how you are using the term. (Nevertheless, I have not looked at your document in any detail.)

Advantages (Over Mainstream Religion)

This strong natural theistic view is immune to:

1. Issues with historicity;
2. Having to depend on historical, Divine Revelation for morality;
3. Having to depend on the passing of tradition onto the next generation;
4. Requiring to accept the writings or reject the whole theory of all the religious scriptures passed down as canon (in whichever religion we are talking about);
5. Having to depend on faith (viz., trust in an authority to verify, at least in part, its position); and
6. Having to accept the Divinity of any given person in order to be saved (such as in Christianity). — Bob Ross

I think 's observation is insightful, as usual. For example, if we have a non-Deistic God who interacts with creation, then it is very intuitive to move into the idea that God has spoken and men have listened (i.e. faith). If God is interacting (and speaking) but these interactions have no special import, then the question arises of why God is bothering to interact. The marriage between Deism and Christianity seems fraught.

Anyway, I hope to have a closer look at your document in the near future.

I see your point, that by labelling X and IBE, underdetermination may not apply. Labeling it the explanation would be underdetermined. — Relativist

Yes, that's right.

But I suggest that in the real world, we operate on beliefs, which are often formed by inferring to the best explanation from the facts at hand (background beliefs will unavoidably affect the analysis). — Relativist

Well, do I believe that every one of my beliefs is "the explanation," or do I believe that some of my beliefs are "the best explanation"? I think we believe that some of our beliefs are only IBEs, and therefore we believe that some of our beliefs are more than IBEs.

Even if we were perfect at this, the resulting beliefs would still be "underdetermined", but ideally they will be our best explanation for the set of information we have. — Relativist

I would basically argue that some theory which is believed to be underdetermined is not believed. So if I think there are only two theories to account for a body of evidence and that both are exactly 50% likely to be true, then I psychologically cannot believe one over the other.

So I think we would need to get more precise on what we mean by "underdetermined." For example, why do you think "the resulting beliefs would still be 'underdetermined'"? Does that mean that the person might change their mind when they reconsider the evidence from a different point of view? If so, then I would say that that possibility to change one's mind (and one's ratio or angle of perspective) is different from one's belief being underdetermined. On Aquinas' view, that form of 'underdetermination' is essential for free will. Apparently it is possible to read the complement of "underdetermination" as fatalism or determinism.

So, we don't "resolve underdetermination tout court," rather, we resolve some specifically pernicious instances of its application. And then, when it comes to scientific theories, the problem of underdetermination is less concerning because our knowledge isn't just a sort of statistical model, which if radically altered, has "remade the world." When we shift paradigms, it isn't that the old world of trees, fire, stars, and sound is revealed to be illusory, and a new socially constructed world has taken its place. We are still dealing with the same actualities as apprehended through new conceptual means. And crucially, while there might be many ways to correctly describe something, these will be isomorphic. When underdetermination becomes more pernicious is when it denies this isomorphism, such that scientific findings become "sociology all the way down" or "power struggles (will to power) all the way down." — Count Timothy von Icarus

Yes, I think that is a good elaboration. :up:

Hence, if we do not like the skeptical conclusions, we should take a look at the epistemic starting points that lead to them.

Indeed, if an epistemology leads to skepticism, that might be a good indication it is inadequate. — Count Timothy von Icarus

Right, and not to get ahead of things, but it is this crucial move that is especially interesting (and is also seen in thinkers like Nathan Jacobs). We eventually come to this point where one must decide whether they prefer the skeptical premises even at the expense of the absurd conclusions. Your "good indication" sums up the complexity of this, because this approach is not demonstrative. That's not necessarily a problem, and perhaps there is no strictly demonstrative alternative, but it is worth noting how the inference at this crucial juncture is rather sui generis (and could even perhaps be construed as coherentist, depending on one's appraisal of the reductio and reductio arguments in general).

To me, that is an insane decision that makes no sense at all. — Athena

Why do you think it's insane? It seems eminently reasonable to me. In fact I can't imagine a real philosophy forum that did not incorporate such a rule. It seems that all of them have.

Because it's a forum for people to talk with other people. — Outlander

Yep. :up: :up:

I'm sort of saying "Well, what if the radical conclusions are true, after all? Maybe it's the realist philosophy of science which is wrong, then" — Moliere

Sure, and I want to see arguments for and against underdetermination, for and against realism. It makes no difference whether Aristotle or Aquinas or Aladdin are the ones who made the good arguments. Focusing on individuals at the expense of the arguments is no help, especially at the very beginning of a thread.

So let's look at your arguments:

Underdetermination is the theory that theories are not determined by the evidence, but rather are chosen in order to organize the evidence, and in some way are a selective pressure on which evidence is relevant to consider. — Moliere

Okay, that is an interesting definition. :up:

1-4 are observations of human beings attempting to generate knowledge which fit with this belief -- basically an IBE, or really just a set of reasons for why I think underdetermination is a good default position. I.e. I don't have a deep quandary with denying causation as a metaphysical reality. That's because causation isn't real but how we decide to organize some body of knowledge.

Closer, or does that just read as more of the same to you? — Moliere

No, I think this is all quite helpful. We now have alternative definitions, arguments, considerations, objections, etc. :up:

(I will come back to this when I have more time. I was mostly trying to expend some effort to try to keep the thread focused on the central theses and the arguments).

That empiricism and academic skepticism died out, in part perhaps because of these arguments, is why St. Thomas doesn't have them as major contenders to rebut in his epoch. — Count Timothy von Icarus

That seems likely. :up:

I am not sure about the rhetorical strategy of continually expressing perplexity about the doctrine you are expounding on or its use by people you are criticizing. I think though that in this case it actually suggests a real confusion it probably doesn't mean to imply. The argument for why form in the intellect (the intellect's move from potency to actuality) cannot be unrelated to its causes comes from the idea that: a. every move from potency to act has a cause in some prior actuality; b. causes cannot be wholly unrelated (i.e. arbitrarily related) to their effects (completely equivocal agents) or else they wouldn't be causes in the first place and what we'd actually have is a spontaneous move from potency to act. Form is just that which makes anything actual to effect anything at all, so form is, in one sense, always present in all causes (granted there are analogical agents). Arguing for this doesn't require question begging and presupposing the doctrine, it requires upstream premises (I see now that Klima appears to have hit on this in more detail). — Count Timothy von Icarus

I think this is good. I don't mean to derail the thread with those papers, but they are related to these central theses of your OP:

1. Things do not happen “for no reason at all.” Things/events have causes. If something is contingent, if it has only potential existence, then some prior actuality must bring it into being. It will not simply snap into being of itself. Our experiences are contingent, thus they must be caused by something that is prior to them.

2. Being is intelligible, and to be is to be intelligible. Every being is something in particular. That is, it has a form, an actuality, that is determinant of what it is (as well as the potential to change, explained by matter). This actuality determines how a thing interacts with everything else, including our sense organs and our intellects. If this was not the case, interactions would be essentially uncaused, and then there would be no reason for them to be one way and not any other (i.e. random). — Count Timothy von Icarus

I'd say that if people want to object to the OP then this is a good place to begin.

At play is the classic clash of metaphysics-first vs epistemology-first paradigms, and it's fairly hard to find a rope such that both sides can grab on and start tugging in different directions. This is actually why I think folks like @Moliere are ultimately tempted to take the shortcut of relativizing Aristotle or Aquinas.

So at the birds-eye level, we have something like:

A. If (1) and (2), then underdetermination is false
B. (1) and (2)
C. Therefore, Underdetermination is false

...that's at least the argument coming from your vantage point. There are arguments for ~C coming from other vantage points above.

Broadly speaking, an argument from underdetermination is one that attempts to show that available evidence is insufficient to determine which of several competing theories is true. That is, many different theories might be able to explain the same evidence, hence any move to choose between theories must be “underdetermined,” i.e., not determined by the evidence. Within the class of such arguments, there are many that go a step further. These will often purport to show that for any body of evidence, there will always be an infinite number of different explanations that are consistent with that evidence. — Count Timothy von Icarus

Regarding the definition of underdetermination, we seem to have at least three options:

• U1. In this particular circumstance, the available evidence is insufficient to determine which of several competing theories is true
• U2. For every body of evidence, there will always be an infinite number of different explanations that are consistent with the evidence
• U3. For every circumstance, the available evidence will be insufficient to determine which of several competing theories is true

I think everyone agrees that (U1) does occur. (U2) may be a point of contention, yet (U3) is probably the point of contention that the average opponent would be more comfortable defending. So perhaps (U3) helps narrow the thesis in question, and if so, then I still think the arguments you have offered are decisive.

If we accept abductive reasoning (inference to best explanation on available evidence - IBE) as leading to rational beliefs, is there really a problem? Such beliefs will, of course, be undertermined but that just means they don't comprise knowledge (in the strict sense).

Epistemology should be of practical use in the world, and in the real world we are nearly always deriving conclusions from limited information. IBEs are the practical ideal. — Relativist

This is a really interesting objection. Is an IBE underdetermined? Remember that the conclusion is not, "X is the explanation," but rather, "X is the best explanation." I actually don't see why underdetermination would need to attend IBEs.

A significant part of this thread will turn on what exactly is meant by "underdetermined."

Is it enough to say

"Modern philosophy has problems. These medieval thinkers didn't have these problems. This is because modern philosophy invented this problem for itself by stripping out all the thoughts which earlier thinkers relied upon in making such inferences. Therefore, we should adopt these earlier approaches, given the incredible progress knowledge has made -- there is a disconnect between ability, and these supposed modern problems that we can pass over by reading the older solutions" ?

Does that demonstrate having read the OP? — Moliere

No, not really. No mention of underdetermintion or realism. You're basically assuming that the OP is about something that it doesn't claim to be about, hence the ad hominem nature. The OP is about underdetermination and realism. That's the core.

My thinking is with respect to underdetermination and its value -- what I read were some solutions to underdetermination based on a generalization of a few select authors rather than what I might say in favor of underdetermination, for instance. So I wanted some sort of reason why these are even appealing at all? — Moliere

So you think Hume, Wittgenstein, Quine, Sextus Empiricus, etc., offer poor arguments for underdetermination? That's intelligible. What is your alternative argument for underdetermination?

For myself I don't feel a deep need to argue for underdetermination because to me it explains why we go through all the hoops we do in making scientific inferences -- we don't just see the object as it is, we frequently make mistakes, and go about looking for reasons to justify our first beliefs while discounting possibilities not on the basis of evidence, but because they do not fit. This is inescapable for any productive thought at all -- but it has the result that we only have a tentative grasp of the whole. — Moliere

Okay, well that's a good start for an argument for underdetermination. :up:

I would want to actually look at some of these premises you are alluding to. For example:

1. We don't just see the object as it is
2. We frequently make mistakes
3. We frequently go about looking for reasons to justify our first beliefs
4. We have only a tentative grasp of the whole
5. Therefore, Underdetermination explains why we go through all the hoops we do in making scientific inferences

I don't see how (5) follows from your premises. Here is something you might want to revisit regarding premises like 1, 2, or 4:

This does not imply that we come to know everything about the actuality of the form. Indeed, we will never know everything about any sort of thing. As Aquinas’ famously put it: “all the efforts of the human intellect cannot exhaust the essence of a single fly.” Nonetheless we know what a fly is. We understand it. It is this phenomenological experience of understanding that is the key datum which epistemology is supposed to explain. — Count Timothy von Icarus

This is something I thought while reading MacIntyre. Yes, I see what you're saying, but like Heidegger you're sort of inventing a whole mindset that is "pre-modern", and justifying it with many quotes -- but at the end of the day if you haven't spoken to people from the pre-modern era then, my brother in christ, you cannot make claims about how pre-modern people think no matter how many texts you read from that era.

It elucidates how we think, but it may not be the panacea of problems contemporary philosophy faces.

It looks soothing -- but ultimately when someone says that if we go back to some ancient or medieval thinker as the person who saw it all I think that we're kind of fibbing to ourselves. — Moliere

You don't seem to be engaging the OP at all.

We're attempting to reconstruct the thoughts of people we can't talk to, yes. — Moliere

Such is philosophy. You do the same thing with the philosophers you appeal to and interpret.

There is lots of substance in the OP. Why not address that substance instead of trying to undercut it with ad hominem gesturing towards Aristotle or Aquinas? Count spent a fair amount of time on this. I would want to honor that.

I want to see some responses that show evidence that the OP was actually read. Your post doesn't manage that. It could be recycled for any Aristotelian-Thomistic OP, regardless of content. Therefore if one reads your post they will not be given any insight into the content of the OP, given that your post in no way reflects the content of the OP. They will only be able to infer that the OP involves the thought of Aristotle and Aquinas.

This is a wonderful essay, eminently relevant. Its work in clearing away canards cannot be overestimated. Its research and accuracy are commendable. It is long yet worthwhile and readable.

Yet if an epistemology results in our having to affirm conclusions that seem prima facie absurd, and if further, it seems to lead towards radical skepticism and epistemological nihilism, or an ever branching fragmentation of disparate “skeptical solutions” and new “anti-realisms,” that might be a good indication that it is simply a bad epistemology. — Count Timothy von Icarus

I recently listened to Nathan Jacobs' podcast on realism and nominalism, which I thought gave a good overview of the territory. I started but have not finished his follow-up podcast, "The Case for Realism," in which he argues for realism and explains why he abandoned nominalism for realism. So far it has been good, and has tracked some of the same points you are making.

Knowledge of trees, an understanding of what a tree is, comes from the presence of this form in our intellect after it has been abstracted from the senses. — Count Timothy von Icarus

There is a good exchange on this point between Robert Pasnau and Gyula Klima, where Pasnau takes up skepticism arguendo against Aristotle and Klima responds.

The result is that the underdetermination of sheer prediction becomes unanswerable, and skepticism reigns.* — Count Timothy von Icarus

I would want to add that the realism quandary is also internal to "predictionism." The one who predicts is attempting to predict ad unum (towards the one, actual, future outcome). Without that future-oriented determinacy—whether actual or theoretical—the "predictionist" cannot function.

I am looking forward to following this thread. I think it will be especially hard to keep it on-topic given that it touches on so many neuralgic subjects which could lead us far afield of the OP.

Edit: with Superman we could also avoid belief and still get an apparent error with "Clark Kent appears on the Daily Planet payroll." — Count Timothy von Icarus

Right, or:

On your reasoning, we can disprove the thesis simply by noting that Superman wears a cape whereas Kent does not. — Leontiskos

I'm sympathetic to most of what you have been saying. But this contradiction can easily be resolved. "Superman" and "Clark Kent" are both names for the same person - but each name is assigned to a different persona. This is not particularly strange - pen names, professional names, character names (Barry Humphries, for example), regal names, baptismal names, adoptive names, married names, aliases of all sorts. — Ludwig V

Well, if the personas have different properties then you have solved the "puzzle."

As it is being trained to complete massive amounts of texts, the model comes to develop latent representations (encoded as the values of billions of contextual embedding stored in the hidden neural network layers) of the beliefs of the authors of the text as well as the features of the human world that those authors are talking about. At some stage, the model comes to be able to accurately impersonate, say, both a misinformed Moon landing hoax theorist and a well informed NASA engineer/historian. However, in order to be able to successfully impersonate both of those people, the model must be able to build a representation of the state of the world that better reflects the knowledge of the engineer than it does the beliefs of the conspiracy theorist. The reason for this is that the beliefs of the conspiracy theorist are more easily predictable in light of the actual facts (known by the engineer/historian) and the additional assumption that they are misguided and misinformed in specific ways than the other way around. In other words, the well informed engineer/historian would be more capable of impersonating a Moon landing hoax theorist in a play than the other way around. He/she would sound plausible to conspiracy theorists in the audience. The opposite isn't true. The misinformed theorists would do a poor job of stating the reasons why we can trust that Americans really landed on the Moon. So, the simple algorithms that trains the model for impersonating proponents of various competing paradigms enable it to highlight the flaws of one paradigm in light of another one. When the model is being fine-tuned, it may be rewarded for favoring some paradigms over others (mainstream medicine over alternative medicines, say) but it retains the latent ability to criticize consensual opinions in the light of heterodox ones and, through suitable prompting, the user can elicit the exercise of those capabilities by the post-trained model. — Pierre-Normand

Thank you again: that is very helpful. As someone who has pondered that general phenomenon, your account makes a lot of sense.

It's interesting that among humans there is another factor which seems to allow the conspiracy theorist to be better informed about the scientific orthodoxy than the layman is informed about the conspiracy theories. This is presumably because the conspiracy theorist more often faces objections to his views (and thus forms counter-arguments), whereas the layman who accepts the reigning orthodoxy will not face objections as often, and therefore will not form counter-arguments and self-reflect on his own reasoning as often. This is perhaps even more obvious when it comes to ideological minorities than conspiracy theorists per se.

My guess is that—supposing this phenomenon does not affect LLMs—the reason it does not affect LLMs is because the LLM has the "time" and "effort" available to expend on the conspiracy theorist, whereas the layman does not. (This gets into the "fairly difficult question" you reference below, namely the manner in which democratic thinking diverges from correct thinking.)

There is both low-level continuity and high-level shift in telos. At the low level, the telos remains accurate next-token prediction, or, more accurately, autoregressive selection. At the high level, there occurs a shift from aimless reproduction of patterns in the training data to, as GPT-5 puts it "assistant policy with H/H/A (helpful/harmless/accurate) goals". How the sense that the model develops of what constitute an accurate response, and of how accuracy is better tracked by some consensual opinions and not others (and sometimes is better tracked by particular minority opinions) is a fairly difficult question. But I think it's an epistemological question that humans also are faced with, and LLMs merely inherit it. — Pierre-Normand

Indeed. Thank you. :up:

I'd suggest that the sheer instrumentality of the "new science" is a major culprit here. It leads to a sort of pride. It's a particularly pernicious pride in that it often masquerades as epistemic humility. Its epistemic bracketing is often an explicit turn towards the creature and the good of the creature without reference to the creator, as if the one could be cut off from the other. "Professing themselves to be wise, they became fools," and exchanged a holistic view for a diabolical process that cuts apart and makes it so that "reason is and ought to be the slave of the passions." — Count Timothy von Icarus

And what's interesting to me is the way that philosophical and cultural degradation is often very subtle, especially at its inception. For example, it is quite difficult to pinpoint where this "proceduralizing" of reason began, even though we can see it fully flowered in the modern period.

Then beginning where we are now, medicinal movement in the proper direction poses the same sort of question. People and especially cultures cannot do a 180° reversal in a day, and usually the reordering must be done via a multitude of small and subtle shifts or re-orderings. It's remarkably difficult for a rationalistic mindset to yield even the smallest concessions (and thus re-orderings) to a more holistic paradigm. It's almost as if the rationalistic context must be abandoned for a time, in much the same way that someone who has developed bad habits of gait should just go swimming for a few hours in an attempt to forget and reset the whole realm of walking.

Merely syllogistic logic cannot deal with modal or other intensional contexts. It treats identity as just another predication. That's one of the reasons it's not much used anymore. — Banno

You're floundering in your continual ignorance of history and philosophy, and you still haven't answered the question at hand. Ironically, the problem with your own view is that it treats identity as just another predication, failing to think through the equivocation between self-identity and other-identity.

Leibniz' whole point was that if you have two things with all the same properties, then you don't have two things. You were mistaken and there is only one thing after all. Thus the "=" on your definition is by definition not a two-place relation. Instead it is a reflexive relation where the object is identical to itself, and where we have mistaken a single object for two different objects. — Leontiskos

Nothing in God's late condemnation of Saul suggests the misrepresentation thesis. — Count Timothy von Icarus

I haven't read much on this point, but it seems to me that the condemnation remains intelligible as long as we don't take the Samuel thesis to an extreme. Saul's actions seem indefensible in general.

So I think this line could be fruitful even if we don't go the route of the "deceiving" thesis:

I remember the writing in bSamuel as brilliant and capturing what can happen even when legitimate prophecy is granted to the crooked timber of humanity.

...

In Torah, you'll hear, e.g., "And God said to Abraham...." In the book of Samuel, this doesn't happen, and instead, it's Samuel telling Saul to put Amalek under the ban. The key here is Samuel. He could be correctly and perfectly conveying God's will, or he could be mistaken, or he could be deceiving. The clarity of Torah, where we see God's words openly dictated, is no longer present in Samuel. — BitconnectCarlos

Meier's Themes and Transformations in Old Testament Prophecy would generally support this thesis, as he argues that the potency, competency, and clarity of prophets gradually diminishes as the Bible draws on.

So true. :fire:

---

I continue to be impressed by the amount of gymnastics — jorndoe

I continue to be impressed by the anti-Trinitarian "mysticism". "It's totally contradictory. I don't have a coherent argument for that conclusion, but just take my word for it!~" Usually one comes to the conclusion that something is contradictory after making a serious argument to that effect. Not on TPF, though. :wink:

Dude, — Banno

On your reasoning, we can disprove the thesis simply by noting that Superman wears a cape whereas Kent does not. Therefore they are not equal or identical. — Leontiskos

You're just kicking the can and avoiding the concrete question. What you mean by "=" is something like, "equivalent with respect to the properties that we take to be relevant," and you haven't the slightest idea of what should count as a relevant or irrelevant property. Else, on your "all properties" account, there are no two things that are equal. Therefore, as I said, your "=" is just a matter of different names for the same thing. Again:

Consider two biconditionals:

SC: The two terms can be substituted salva veritate within this context ↔ The two terms are equivalent within this context
SA: The two terms can be substituted salva veritate in every context ↔ The two terms are equivalent in every context (i.e. the two terms are absolutely identical)

Both of these biconditionals are true, but this is the argumentation that leverages SA:

i. [Claim that two terms can be substituted in every context]
ii. [Identify a context in which the two terms cannot be substituted]
iii. Draw a reductio of some kind

For example:

1. "Superman" = "Clark Kent."
2. Lois believes that Superman can fly.
3. ∴ Lois believes that Clark Kent can fly.

As I pointed out above, (1) is false, but it is false in a very deep sense. This is because SA is a linguistic impossibility, and therefore to stipulate that some pair of terms satisfies SA is to stipulate a linguistic impossibility. It’s therefore no surprise that one can always find a context in which the two terms cannot be substituted once one moves out into the real world. — Leontiskos

SA is a linguistic impossibility. Leibniz' whole point was that if you have two things with all the same properties, then you don't have two things. You were mistaken and there is only one thing after all. Thus the "=" on your definition is by definition not a two-place relation. Instead it is a reflexive relation where the object is identical to itself, and where we have mistaken a single object for two different objects. It is epistemology trying to disguise itself and pass for a static proposition. It is a conflation of reference and referent.

"Superman = Clark Kent" is logically presupposing both that there are two things being related, and that there are not two things but only one thing. It's that inherent contradiction that is the problem, and which is so bound up in your own thought.

I apologize: I was not understanding you before. I thought you were referring to demonic possession. Indeed, I agree that it is much more questionable if demonic hybrids would have rights. — Bob Ross

That's alright - it's an understandable assumption. At this point we are knee-deep in obscura. :smile:
For example, according to the secondary literature the demons that Jesus casts out were originally spawned by groups like the Amalekites, and roamed the Earth looking for hosts after being killed by the Israelites.

Could God wipe them out justly? I don’t know, but it would definitely violate the rationale I gave above for rights. — Bob Ross

That's a fair argument you give. What's interesting is that when Jesus encounters these demons that—according to the secondary literature—originally came from groups like the Amalekites, they say things like this:

And behold, they cried out, “What have you to do with us, O Son of God? Have you come here to torment us before the time?” — Matthew 8:29 (RSV)

The backstory here is that in his mercy and providence, God has allowed such beings to continue to exist on Earth until "the time," namely the end times. So oddly enough, there is a respect even for demons built into the narratives. Jesus even accedes to their request in v. 32.

Yes, but no one that objects with those to me (so far) has ever coherently defined what ‘murder’ is. Like I said, that view may be internally coherent in some theory; but it isn’t coherent with the idea of rights I expounded above. Do you have a different definition of murder that you prefer such that God and the Angel of Death are not committing murder?

My definition, to recap, is that murder is the direct intentional killing of a person. — Bob Ross

Yes, and it's fair enough that you would press your point. Let's try to understand the logic a bit. First, your argument, which of course presupposes that murder is impermissible:

1. Murder is the direct intentional killing of an [innocent] person
2. The Angel of Death intentionally kills the innocent Amalekite infant
3. Therefore, the Angel of Death is a murderer

And then the reductio I mentioned (although I will not here present it as a reductio):

4. It is the Angel of Death's job to take life
5. It is not impermissible to do one's job
6. Therefore, the Angel of Death is not a murderer

This is the case where there is a logical standoff between two contradictory conclusions, and yet there is no attempt to formally invalidate the opposing argument. Formal reductios also function in precisely this way. If we have only these two arguments, then one must simply weigh them and decide which is stronger.

Digging deeper, (4) and (5) have to do with the idea that death is inevitable, and that for a person to die is not inherently unjust. This opens up the can of worms of the metaphysics and ethics of death, and the adjacent can of worms is the question of God's sovereignty within which question is the matter of whether God is responsible for death (or whether God "directly intends" the fact of natural death).

So this all gets complicated quickly, and therefore it is hard to try to capture the various complexities with a syllogism or two. For example, if everything that occurs is allowed by God to occur, and if this allowance counts as an intentional bringing-about, then it follows that everyone who dies is murdered. The reductio in this case lies in the idea that murder and death are two different things. Note too that we are wrestling with precisely the same issue that the Hebrews wrestled with in trying to understand God's sovereignty and providence (in, for example, hardening or not-hardening Pharaoh's heart).

Interesting. It seems like Fr. Stephen is taking a more spiritual approach to the theology and the Bible (going back to the beginning of our conversation). — Bob Ross

I wouldn't say that he takes a more spiritual or metaphorical approach to theology and the Bible, but I can see how this video in particular might produce that idea.

His critique is fair insofar that systematizing is can go too far and systematize for the sole sake of doing so (e.g., Kant); but I wonder how valid this critique really is: he seems to just have given up on striving towards perfect knowledge. It seems like systematic knowledge is just the attempt at, or aspiration towards, complete knowledge. Should we really give that up? What do we have left after doing so? — Bob Ross

This is a really interesting and complicated topic, but I will try to say a few things.

In general we recognize that one must collect the data before they form their theory or propose their thesis. We also recognize that if a theory is invalidated by data, we have to accept that rather than stubbornly cling to the theory while ignoring the data. I think De Young is saying that a lot of people have over-simple theories that run into problems when deeper and broader datasets are encountered. For example, I am told that there is a fun documentary on the Super Smash Bros video game, which follows different groups of people who thought they were the best and had mastered the game, only to find that others were much better (and that South Koreans are often elite in such matters).

It's something like that: you thought you understood it until you understand that you don't. That is Socrates' virtue: an understanding of his own limitations and ignorance. De Young is saying that when it comes to God this phenomenon gets taken to a whole new level (cf. Isaiah 55:8-9).

At the same time there is the danger of falling into the other extreme, which is what I think you are speaking to. There is the danger of skepticism or despair of knowledge altogether. There is the danger of theological voluntarism where God becomes wholly inscrutable. Yet what happens when one settles into a deep tradition such as Christianity, is that they settle into the habit of finding they were mistaken and thus being prepared to see how they are currently mistaken. This creates an openness to a reality beyond them (and this same phenomenon occurs when someone takes on a teacher, acknowledging that they have much to learn). I want to say that this humble stance towards reality and God is incredibly important, even if one rejects Christianity. We can of course reject things, but (please God) we should never find ourselves in a place where a self-confidence has closed us off to reality or to that which transcends our own capacities.

That must mean there is something objective and particular about the concept of the Trinity — Fire Ologist

I think this is right. The "sensus fidelium" could not exist if what is agreed upon were truly incoherent.

That was a very good bringing-together of various different strands of the thread. :up:
I was revisiting Damascene's exposition of the faith to see how much more accessible it is than Aquinas. It is certainly more accessible, but perhaps still not accessible enough for what this thread would require.

Wholly instrumental analytic reason is in a sense diabolical (in both its original and current sense). — Count Timothy von Icarus

Agreed. And we can hold that such an approach is diabolical while also maintaining that it need not be explicitly atheistic (for example). The issue has to do with a closed-off-ness to both analogical reasoning and transcendence.

-

Let's just leave it at this: on it's face, the Catholic Trinity appears to be contradictory. — frank

If there is a contradiction, then present the argument for that conclusion. As I noted earlier, internecine differences over the Trinity do not turn on the question of contradiction. They turn on the question of consistency with Scripture, the Fathers, or other such sources.

Thank you, that is very helpful! Let me ask a few follow-up questions.

As a result, the base-model has the latent ability to express any of the wide range of intelligible opinions that an author of some piece of the training data might have produced, and has no proclivity to adjudicate between them. — Pierre-Normand

Isn't it true that the opinions of the author of some piece of training data will converge in some ways and diverge in others? For example, the opinions might converge on the idea that slavery is wrong but diverge on the question of who will be the Governor of Nevada in 2032. If that is right, then how does the LLM handle each case, and how does one know when the opinions are converging and when they are diverging? Similarly, [what] criteria does the LLM use to decide when to present its answer as a mere opinion, and when to present its answer with more certitude?

During post-training, the model's weights are reconfigured through reinforcement learning in order to fit the schema USER: <query>, ASSISTANT: <response>, USER: <follow up question>, etc. and the models responses that are deemed best in accordance with predetermined criteria (usefulness, harmlessness, accuracy, etc.) are reinforced by human evaluators of by a reward model trained by human evaluators. Some political biases may arise from this process rather than from the consensual or majority opinions present in the training data. But it is also a process by means of which the opinions expressed by the model come to be pegged rather closely to the inferred opinions of the user just because such responses tend to be deemed by evaluators to be more useful or accurate. (Some degree of reward-hacking sometimes is going on at this stage). — Pierre-Normand

Great.

So suppose the LLM's response is an output, and there are various inputs that inform that output. I am wondering which inputs are stable and which inputs are variable. For example, the "post-training" that you describe is a variable input which varies with user decisions. The "predetermined criteria" that you describe is a stable input that does not change apart from things like software updates or "backend" tinkering. The dataset that the LLM is trained on is a variable input insofar as one is allowed to do the training themselves.

I am ultimately wondering about the telos of the LLM. For example, if the LLM is designed to be agreeable, informative, and adaptive, we might say that its telos is to mimic an agreeable and intelligent person who is familiar with all of the data that the LLM has been trained on. We might say that post-training modifies the "personality" of the LLM to accord with those users it has interacted with, thus giving special weight to the interests and goals of such users. Obviously different LLMs will have a different telos, but are there some overarching generalities to be had? The other caveat here is that my question may be incoherent if the base model and the post-trained model have starkly different teloi, with no significant continuity.

It's more akin to a rational reconstruction of the opinions that the model has learned to produce under the constraints that this response would likely be deemed by the user to be useful, cogent and accurate. Actual cogency and accuracy are achieved with some reliability when, as often is the case, the most plausible sounding answer (as the specific user would evaluate it) is the most plausible answer. — Pierre-Normand

Okay, interesting. :up:

(I also read through some of your GPT links. :up:)

Googling "God: Multiple persons sharing one being" returns the Trinity. — Banno

Is that really a surprise?

Banno is a case in point for the future in which LLM-based arguments become synonymous with a kind of self-fulfilling prophecy bias. What Narcissus would have given for an LLM to reflect back to him his own prejudices!

↪frank But I think what I've said in the above posts acknowledges all of that. I said:

So two men both 'participate' in the form 'man' even though they are numerically different men. — Wayfarer

One (rather limited) way of approaching the Trinity is as a mean between the extreme of a strong emphasis on the persons (which moves in the direction of polytheism) and the extreme of a strong emphasis on the unification of the divine nature (which moves in the direction of emphasizing the divine nature at the expense of the personal distinctions).

So on this scheme we're sailing down a river where the north shore is polytheism, the south shore is ousia-overemphasis, and we want to stay in the middle of the river and avoid crashing into either shore. As with all areas of subtle philosophy, overcorrection is a constant danger.

Banno presented " an unlikely strawman that very little attention is [historically] paid to the idea at all." You responded by pointing to the idea that "is God" is indicating the predication of a nature, and that is definitely the right response to Banno's odd transitivity argument. You gave an example of two men who participate in the form of 'man' (human), which is also a helpful illustration. The quibble against your example is that the persons of the Trinity are not separated from each other in the way that human persons are separated from each other, and that if for some reason we take your example to be identical to the Trinity, then it fails: it veers too close to the north shore. This objection is intelligible, but I don't see any reason to assume that you were offering the example as something more than an analogy.

(Another way of viewing that objection is as a utilization of Euthyphro-like reasoning against the reification of the ousia.)

The danger for the Catholic is polytheism. — Banno

No, Catholicism (and Western Christianity in general) has always veered closer to the south shore. Polytheism is the danger for Eastern Christianity.

When Catholics say the Father is God, they are not predicating. They aren't saying God is a category the Father belongs to. It's an identity statement. The Father is not a section of God. The Father is fully God. Whatever God is, the Father is equal to that. — frank

No, this is not right. I would go back to my posts where I quoted the Catechism of the Catholic Church. We can say that the Father is God (in the Triune sense), but by that we include the Son and the Spirit with the Father, for they are never apart (except notionally, in the single notional case where an exclusive relation of origin is being considered). More commonly, we would say that the Father is divine (and this is a matter of ousia).

Regarding my post to you about John 1:1, there is an ancient sense in which "God" ("the god") is used hypostatically to refer to the Father, but the semantics of the hypostatic use and the Triune use are distinct.

Christianity is the most ideologically dynamic of all the global religions because it's a fusion of several different sets of cultural outlooks and values. — frank

Perhaps this is most true of Christianity given its geographic sprawl, but it is also true of many other religions to a lesser extent.

Kastrup uses this as a metaphor for the relationship between individual minds and what he calls “mind at large.” Just as each dissociated identity experiences itself as a separate person, we experience ourselves as separate individuals—when, in his view, we are all expressions of the same underlying mind manifesting in different ways. — Wayfarer

In a more individual way Trinitarian thought is often applied analogously to psychological health. When the various aspects of one's personality become dissociated, one becomes mad, corrupt, divided, schizophrenic, etc. When the various aspects of one's personality enter into a symbiotic and fruitful union, one achieves psychological health. Monomania would be the case where only one aspect of one's personality is allowed to continue existing.

And I'll note that you've failed to answer the simple question, "What does <Superman = Clark Kent> mean?," three times now.

More of the same from Banno.

Why not just admit what I've said from the start: that you don't know what you mean when you say things like, "Superman = Clark Kent"?

On your reasoning, we can disprove the thesis simply by noting that Superman wears a cape whereas Kent does not. Therefore they are not equal or identical.

This is not a complete account, but it'll do. — Banno

So again, what does <Superman = Clark Kent> mean? And is the "binary predicate" true or false in this case?

(I of course responded to your confusions yesterday.)

"=" is very well defined in both maths and logic — Banno

Then give your account of what <Superman = Clark Kent> means.

I only grant that it is well-defined in mathematics.

Analytics like Banno seldom have any idea what they are doing when they say, "x = y," as they assume that anything can be placed into that form. They don't recognize the mathematical context and the single genus of the relata that their formulation takes for granted. — Leontiskos

That conclusion (not premise) could only be made by someone who knew both the differences and sameness between what is a “Clark” and what is a “Superman”. — Fire Ologist

Right. In order for (1) to avoid tautology there must be rational movement, and this requires some difference between the two relata.

And of course I don't mean that we can't use (1) as a premise, but rather that we must be prepared to give an argument for such premises. Such premises are not self-evident.

P1: X = Y
P2: Z is ready enough to say "X can fly."
P3: Therefore, Z is ready enough to say "Y can fly."

I don’t think this apparent controversy is about an apparent flaw in the notion “X = Y”, but from the insertion of the “Z is ready to say that…”. Z’s belief creates a new context in which we must redefine X and Y. So we can’t substitute the use of either X or Y from P1, in any sentence following P2; P2 has redefined X and Y according to Z’s belief. — Fire Ologist

Yeah, I think you're right about this. P3 requires a premise about whether X = Y for Z.

But drumming my point, we could also scrutinize P1. What does P1 mean? In the Superman case it is supposed to mean that one and the same thing goes by two different names. Is that what it means in a mathematical context? I'd say the fact that we don't really know what we are saying with (1) is significant. Given that there are so many multivalent meanings to P1, it is itself a kind of analogical claim. Presumably disambiguating P1 would shed light on P3.

How did I end up analogizing the Trinity to a single human person, and it jibes with Aquinas, but I didn’t go to Aquinas? Incoherence in the notion of a ‘Trinity’ would make this an utter accident. — Fire Ologist

Yes, and what's interesting here is that the development of the concept of personhood had a great deal to do with Trinitarian theology. The precision that we now have around the word "person" did not exist in the 4th century. Theater and Trinitarian thought were two of the principle ways that the concept was developed.

There is a similar way in which someone might think that a cardinal (bird) looks like a Roman Catholic Cardinal, or that cappuccino looks like a Capuchin's habit. In fact the bird and the coffee were named after Cardinals and Capuchins, and so the causality is reversed.

This all helps give the lie to the idea that religious thinking is somehow private or irrelevant. Religious thinking forms the basis for much of our current thought and language.