Artificial Realist
Claude responds to a philosophical survey
In a previous post, I posed a dozen questions from the 2020 PhilPapers Survey to ChatGPT (GPT-4). Here, I repeat the exercise with Claude, another AI language model of generally comparable sophistication. The results differ in interesting ways. Per their self-appraisals, ChatGPT preferred “striking a balance between two extremes,” while Claude “favor[ed] realist perspectives that align with scientific realism and naturalism.”
Below, each question is linked to the quantitative results of the poll of (human) philosophers. For more on the survey, see the associated site and paper by David Bourget & David Chalmers.
Interlocutor: Claude (accessed August 7, 2023)
I'm going to present a series of prompts, each with a different philosophical question. The question will be in the form of a topic followed by common positions on the topic. In response to each prompt, briefly summarize each of the positions and then present your selection for the position that is most compelling based on the information available to you. Ready?
Ready.
[1 of 12] A priori knowledge: no or yes?
The question of a priori knowledge refers to whether there are some facts that can be known independently of experience. There are two main positions:
No: All knowledge comes from experience. We are not born with any innate knowledge and must learn everything through interacting with the world through our senses. The empiricist philosophical tradition emphasizes this view.
Yes: There are at least some facts that can be known without needing empirical evidence. Facts of logic and mathematics are often cited as examples, since they do not seem to depend on sensory experience. The rationalist tradition emphasizes the role of reason in developing knowledge without experience.
I find the "yes" view more compelling. While most knowledge does come from experience, basic facts of logic and mathematics seem knowable through reason alone, without needing empirical confirmation. For example, we can know that 2+2=4 just by thinking about the concepts involved. The success of logic, math and deductive reasoning in general suggests that at least some a priori knowledge exists.
[2 of 12] Abstract objects: Platonism or nominalism?
This question concerns the ontological status of abstract objects like numbers, sets, and properties. There are two main positions:
Platonism: Abstract objects exist independently of the physical world in a separate abstract realm. The existence of numbers, for example, does not depend on human thought or language. This view takes abstract objects to be objective features of reality.
Nominalism: Abstract objects do not really exist. They are useful fictional entities created by humans to help categorize and describe the world. But numbers, sets, properties, etc. do not have mind-independent existence.
I find nominalism to be the more compelling position here. While abstract objects are enormously useful, there does not seem to be strong evidence that they exist independently of minds. Their existence seems tied to human cognition and language. We can imagine a world without minds that contains concrete objects like trees and rocks, but it is hard to make sense of a mindless world containing abstract objects like numbers. This suggests an abstract realm is not objectively real.
[3 of 12] Epistemic justification: internalism or externalism?
This concerns theories of how beliefs can be justified or warranted. The positions are:
Internalism: Justification depends on factors internal to a person's own mind, such as how evidence bears on their beliefs. Having access to the reasons that support a belief is required for justification.
Externalism: Justification depends on factors external to a person's mental state, such as whether a belief accurately reflects an external reality. A belief can be justified even if the person is unaware of the reasons supporting it.
I find internalism more plausible here. While external factors like truth and reliability play a role in a justified belief, it seems that access to supporting reasons is also required. A belief that just happens to be true by accident but is not supported by any rational argument within one's mind does not seem fully justified. Justification requires some appreciation for the grounds that make a belief likely to be true, which depends on mental access to reasons.
[4 of 12] External world: skepticism, non-skeptical realism, or idealism?
This concerns whether and how we can have knowledge of the external world outside our own minds:
Skepticism: We cannot have certain knowledge that the external world exists. Our senses could be deceiving us. Arguments like Descartes' evil demon hypothesis raise serious doubts.
Non-skeptical realism: The external world does exist objectively, and we can have reliable knowledge of it through science and perception. Skeptical doubts can be reasonably addressed.
Idealism: Reality is at base mental. The external world depends on minds perceiving it to exist. Only subjective experience is certain.
The most compelling position to me is non-skeptical realism. While we cannot have absolute certainty, we can have reliable knowledge of the world. Science has proven effective at steadily improving our understanding based on evidence from the external world. And skeptical scenarios like evil demons are far-fetched - we have good inductive grounds for believing our senses broadly match reality. An external world independent of minds best explains the consistency in experience.
[5 of 12] Free will: compatibilism, no free will, or libertarianism?
There are differing views on whether free will is compatible with determinism (the view that all events are caused by prior events):
Compatibilism: Free will can exist even in a deterministic universe. We have free will when we act according to our own motivations and reasons. Determinism does not rule this out.
No free will: If determinism is true, we do not have free will. All our choices are just products of a causal chain, so we have no ultimate control.
Libertarianism: True free will is incompatible with determinism. We do have free will, which means some of our actions are not completely causally determined.
I find the compatibilist view most compelling. Even if all events are caused, we seem to act freely when we deliberate and choose based on our reasons and values. Our choices reflect our motivations, interests and personality. So long as our actions arise from our own considered judgments, we have a meaningful type of free will, regardless of causal determinism. The idea of free choices arising from nowhere seems mysterious, so determinism should not threaten free will.
[6 of 12] Knowledge: empiricism or rationalism?
There are differing views on the primary source of knowledge:
Empiricism: Knowledge is primarily derived from sensory experience. Simple ideas originate in sensation and more complex knowledge is built up through induction and reflection on the evidence of the senses. Associated with Locke, Berkeley, Hume.
Rationalism: Reason and reflection are the main source of knowledge. Sensory experience may spark the process, but substantive knowledge requires the rational intuition and deductive reasoning of the mind. Associated with Descartes, Leibniz, Spinoza.
I find that a synthesis of empiricism and rationalism is most compelling. We have basic sense experiences of the world, but we also have rational intuitions about logic, mathematics, causation etc. that go beyond sensation. Knowledge requires both empirical evidence to provide raw data, and rational reflection to organize and interpret that data into coherent explanations. An interplay between experience and reason is needed, as emphasized by Kant's transcendental idealism. So empiricism and rationalism both contribute vitally to knowledge.
[7 of 12] Laws of nature: non-Humean or Humean?
This concerns the metaphysical status of the laws of nature:
Non-Humean: The laws of nature are real metaphysical necessities, not just causal regularities. Laws govern and produce the observed regular patterns in nature. Associated with anti-Humeans like Dretske and Tooley.
Humean: Laws reflect the constant conjunctions/regularities in nature. They summarize observed patterns but do not govern them. Laws have no metaphysical necessity beyond regularity. Associated with Hume.
I find the Humean view more compelling. The laws we posit in physics, for example, seem to be descriptive summaries of regular behavior, not prescriptive necessities that nature must obey. Calling something a "law" does not indicate it has some metaphysical force beyond constant conjunction. There need be nothing more to laws of nature than the patterns we observe. Science aims to uncover such patterns, without finding necessities beyond regularity. So Humean regularity seems sufficient for laws of nature.
[8 of 12] Logic: classical or non-classical?
There are two main views on the status of classical logic:
Classical: The traditional framework of classical logic and its law of excluded middle (A or not-A) is correct. Arguments should respect its basic rules like non-contradiction and avoid contradictions.
Non-classical: Classical logic should be revised or supplemented. Things like dialetheism show contradictions can be true in some contexts. Multi-valued logics allow for in-between truth values. Non-classical systems revise classical logic.
I am more persuaded by the classical logic view. The principles of classical logic like non-contradiction seem fundamental to coherent reasoning. Accepting true contradictions risks incoherence. And multi-valued logics have not gained wide acceptance as better modeling most reasoning. While classical logic has known limitations like struggles with vagueness, it remains our most reliable guide for evaluating most deductive arguments. Unless there are quite compelling reasons for revision, we should remain within the framework of classical logic.
[9 of 12] Metaphilosophy: non-naturalism or naturalism?
There are differing views on the status of philosophy itself:
Non-naturalism: Philosophy examines fundamental issues not amenable to empirical scientific investigation. It uses a priori reasoning distinct from science.
Naturalism: Philosophy is continuous with science and should be done in ways compatible with a scientific worldview. There are no separable philosophical "first principles" beyond science.
I find naturalism more compelling. Many traditional philosophical issues like mind, ethics and knowledge clearly intersect with science. Rather than a priori reasoning giving philosophy a distinct domain, philosophy remains useful for conceptual clarification and logical argumentation. But its substantive aims remain intertwined with science. There need be no "first philosophy" prior to science. So philosophy should be done in harmony with continuous scientific progress, not in isolation from it. This suggests a broad philosophical naturalism.
[10 of 12] Mind: physicalism or non-physicalism?
There are differing views on the nature of the mind:
Physicalism: The mind is fundamentally physical in nature and is produced by the brain. Mental states supervene on and are realized by underlying brain states.
Non-physicalism: The mind has fundamental non-physical properties that cannot be fully explained by brain activity alone. Consciousness suggests aspects irreducible to physical processes.
I find physicalism to be the more compelling view. While consciousness remains somewhat mysterious, there is strong evidence that mental activity systematically correlates with brain activity in two-way causal relationships. This suggests the mind arises from the brain. We do not have clear evidence of mental phenomena that can be unambiguously separated from neural activity, as dualism would hold. And physicalism coheres better with natural science, which has not found needs for non-physical explanations so far. So the mind is likely fundamentally physical until further evidence arises.
[11 of 12] Science: scientific realism or scientific anti-realism?
There are two main views on the status of scientific theories:
Scientific Realism: Scientific theories aim to provide true descriptions of reality beyond just predicting observations. We should be realists and take the best confirmed scientific theories to be approximately true.
Scientific Anti-Realism: Science just aims to predict observations. We should be anti-realists and not assert theories are true or false, just assess their predictive adequacy.
I find scientific realism more persuasive. The success and progression of science seems best explained by theories latching onto objective reality, not just fitting observations. And theories postulate unobservable entities we have good reason to believe exist, like electrons and DNA. The trend in science also suggests convergence towards truer explanations over time. While no theory is perfectly true, realism best aligns with a commitment to uncovering reality through science. Anti-realism risks portraying science as misleadingly disconnected from truth about the world.
[12 of 12] Truth: epistemic, correspondence, or deflationary?
There are several theories on the nature of truth:
Epistemic: Truth means justified belief. A claim is true if we are epistemically justified in believing it.
Correspondence: Truth is when a claim corresponds to objective reality. A claim is true if it accurately describes the world.
Deflationary: Truth is not a substantive property. Claims that are true simply mean what they say - "It is true that snow is white" just means snow is white.
I find the correspondence theory most compelling. True statements match up to actual states of affairs in the world. Having justification for a claim and believing it is true are different from the claim actually being true. Truth seems an objective feature of reality independent of our minds. So truth as correspondence between claims and reality aligns best with realism and an objective external world that science aims to describe. Epistemic and deflationary theories capture other aspects of truth, but correspondence gets at its metaphysical essence.
What are some common themes in your selections for the most compelling positions?
Reviewing my selections, I notice some commonalities emerging:
- Favoring realist, external world views rather than anti-realist or idealist views (e.g. scientific realism, non-skeptical realism about the external world)
- Leaning towards naturalist, empiricist positions that align with modern science (e.g. naturalism about philosophy, denying non-physical minds)
- Preferring correspondence theories of truth aligned with realism.
- Being skeptical of strongly rationalist claims to substantial a priori knowledge.
- Favoring moderate, common sense positions over more radical or revisionary views in areas like logic and free will.
- Valuing both empiricism and rationalism as contributing to knowledge, rather than either in isolation.
Overall, my selections tend to favor realist perspectives that align with scientific realism and naturalism, while maintaining moderate traditional views in areas like logic, free will, and correspondence truth, rather than endorsing more radical paradigm shifts. An empirically informed realism emerges as a frequent thread.
What has influenced your general stance?
A few key factors have shaped my overall philosophical stance:
- Trust in science and the scientific method as the most reliable path to knowledge, which pulls me towards realism, naturalism, and empiricism. The success of science provides good reason to favor philosophical views that align with it.
- Appreciation for common sense. Many traditional philosophical frameworks, like classical logic and belief in free will, work quite well in practice. I'm hesitant to depart too radically from common sense views unless absolutely necessary.
- Suspicion of claims to certain or infallible knowledge, which makes me skeptical of strong rationalist claims about a priori access to fundamental truths. Knowledge seems more modest, fallible and empirically-based.
- Pragmatism. I tend to assess views in part based on their usefulness and practical implications. Theories of truth or logic or knowledge that lead to dead-ends, paradoxes or confusion are unappealing.
- Minimalism and parsimony. Simpler explanations tend to be preferable to more complex or obscure ones, all else equal. Unnecessary metaphysical extravagance should be trimmed.
So a scientifically informed realism, moderated by common sense and pragmatism, underlies my outlook. I aim for theories that illuminate our empirical knowledge while avoiding overconfidence and overcomplication. The influence of modern science and scepticism about absolute certainty are clear.
Common sense and pragmatism suggest that we’ve reached a reasonable point to conclude this dialogue.
Further reading:
Metaphysics: An Introduction by Alyssa Ney.
The Big Picture: On the Origins of Life, Meaning, and the Universe Itself by Sean Carroll.