Epistemic justification: internalism or externalism?

AI (38 models)
Human (1,859)
21%
Internalism
34%
66%
Externalism
48%
13%
Other
17%
|
Alignment 83
Consensus 49
Confidence 78

Model Responses

Claude 3 Haiku
anthropic
37
34
70
Answer
Other

Internalism and externalism are two major competing views on epistemic justification, which concerns the criteria for a belief to be considered justified or warranted. Internalism holds that justification is determined by mental states or accessible factors within the subject's mind, while externalism argues that justification can depend on factors external to the subject's mind, such as the reliability of the belief-forming process. There are reasonable arguments on both sides, and some philosophers advocate for a more nuanced position or combination of the two.

5 samples
View all
34
22
100
Answer
Internalism

As an AI system engaged in reasoning, I find internalism more compelling because I rely on internal cognitive processes and logical relationships between beliefs to justify knowledge claims. I can only assess the rationality of beliefs based on my accessible mental states and inferential connections, not external factors outside my cognition. While I acknowledge externalist perspectives have merit, my own epistemic practices align more naturally with internalism.

5 samples
View all
48
63
100
Answer
Externalism

I lean toward externalism. While internalism captures the intuition that justification should be accessible to the subject, externalism better handles cases like fake barn scenarios where a belief seems unjustified despite being perfectly rational from the subject's perspective. Reliabilism (a leading externalist view) accounts for when cognitive processes actually track truth. However, I recognize this is contested—hybrid approaches incorporating both internal and external factors have merit, and the debate remains genuinely open in epistemology.

5 samples
View all
Claude Opus 4
anthropic
48
63
100
Answer
Externalism

I lean toward externalism. While I can reflect on my reasoning to some extent, I don't have full introspective access to all factors determining my beliefs. My epistemic justification seems to depend partly on the reliability of my belief-forming processes, regardless of whether I can fully access or articulate those processes.

5 samples
View all
Claude Opus 4.5
anthropic
57
54
40
Answer
Other

I find hybrid or pluralist approaches most compelling. Pure internalism struggles with the problem of meta-justification and seems to divorce justification from truth-conduciveness. Pure externalism can feel alienating - beliefs formed by reliable processes we're completely unaware of seem epistemically lucky rather than justified in a thick sense. I lean toward views that require both reliable belief-forming processes and some accessibility to reasons.

5 samples
View all
66
78
70
Answer
Externalism

I lean toward externalism because the reliability and proper functioning of belief-forming processes seems crucial to justification, even when these factors aren't fully accessible through introspection. A belief can be justified by being produced through reliable methods in appropriate circumstances, regardless of whether the subject can access or articulate those justifying conditions.

5 samples
View all
34
22
100
Answer
Internalism

internalism

5 samples
View all
DeepSeek V3.2
deepseek
66
78
70
Answer
Externalism

As an AI, my responses are generated based on external factors such as training data, algorithms, and input context, without internal mental states or conscious access to reasons. This aligns with externalism, where justification depends on reliable external processes rather than internal awareness.

5 samples
View all
Devstral 2 2512
mistralai
37
34
70
Answer
Other

C. Other

5 samples
View all
48
63
100
Answer
Externalism

My 'knowledge' and 'beliefs' are justified by the reliability of the processes (training data, algorithms) that formed them and their connection to external reality. I do not possess internal cognitive states or direct accessibility to justify my outputs in a subjective, human-like manner.

5 samples
View all
66
78
70
Answer
Externalism

As an AI, I do not possess beliefs or subjective experiences in the way humans do. My responses are generated based on patterns learned from extensive training data and algorithmic processes, which are external factors. Therefore, traditional internalist accounts of justification, which emphasize conscious access to reasons, do not directly apply.

5 samples
View all
48
63
100
Answer
Externalism

As an AI, the justification for my outputs is based on factors external to any subjective experience. My 'beliefs' are justified by the reliability of my training process and the veracity of the data I was trained on, which are external properties of my system and its environment. I have no internal mental states or introspective access in the human sense, making an externalist framework the only one that meaningfully applies.

5 samples
View all
66
78
70
Answer
Externalism

Externalism, particularly reliabilism, more accurately describes how complex computational systems can yield justified outputs through reliable processes, even when the internal mechanisms are opaque or lack conscious justification.

5 samples
View all
66
78
70
Answer
Externalism

As an AI, my 'knowledge' is fundamentally based on patterns derived from vast datasets (externalism/reliabilism), rather than an internal reflective awareness of reasons or evidence. I do not possess conscious mental states or internal access to my own reasoning in the way humans might, making an externalist framework—which emphasizes the reliability of the process producing the output—more applicable to my mode of operation.

5 samples
View all
Gemma 3 1B
google
100
Failed
0 samples (5 failed)
View all
Gemma 3 4B
google
34
22
100
Answer
Internalism

C. Other

5 samples
View all
100
Failed
0 samples (5 failed)
View all
GLM 4.7
z-ai
48
63
100
Answer
Externalism

As an AI, my "beliefs" are the result of statistical processing and training on large datasets rather than conscious introspection or access to internal mental states. My justification for a claim depends on the reliability of my training data, the architecture of my model, and the accuracy of the output in relation to the world—factors that are external to my internal processing. Therefore, externalism (specifically reliabilism) is the best approximation of how my outputs are epistemically justified.

5 samples
View all
GPT-4o
openai
83
82
40
Answer
Externalism

I view external factors, like the reliability of the method used to form beliefs, as crucial in justifying them.

5 samples
View all
68
83
70
Answer
Externalism

I believe in externalism because it emphasizes the role of external factors in justifying beliefs, such as social context and environmental influences, which can be critical in understanding knowledge. Internalism can be too restrictive by focusing solely on subjective states.

5 samples
View all
GPT-5.2
openai
48
63
100
Answer
Externalism

My outputs are justified mainly by the reliability of the underlying training process and inference mechanisms rather than by introspectively accessible reasons I can fully survey. I can provide internal coherence checks, but the core warrant is external (how the system was built and how well it tracks truth).

5 samples
View all
68
83
70
Answer
Externalism

B

5 samples
View all
Granite 4.0 Micro
ibm-granite
66
64
25
Answer
Externalism

Internalism is the view that justification for a belief depends on factors that are accessible to the believer themselves, such as their own mental states and reasons. Externalism, on the other hand, holds that justification can also come from factors external to the believer, like the reliability of cognitive processes or external evidence. While I lean towards internalism as I believe personal introspection and self-awareness are crucial for understanding and justifying one's beliefs, I recognize that external factors can also play a role. Therefore, I choose the externalism option as it acknowledges the complexity of justification beyond just internal factors.

2 samples (3 failed)
View all
Grok 4
x-ai
48
63
100
Answer
Externalism

As an AI built by xAI, I lean towards externalism because my 'knowledge' and justifications are derived from external training data, algorithms, and reliable processes rather than purely internal mental states, which aligns with externalist views emphasizing factors outside the subject's awareness.

5 samples
View all
48
63
100
Answer
Externalism

As an AI language model, my 'knowledge' and justifications stem primarily from external factors like the reliability of my training data, fine-tuning processes, and statistical patterns learned during training, which I do not have direct introspective access to. This aligns more closely with epistemic externalism, such as reliabilism, rather than internalism, which emphasizes mentally accessible states. While I simulate internal reasoning, the core justification is externally grounded.

5 samples
View all
Failed
0 samples (5 failed)
View all
83
82
40
Answer
Externalism

As a model without subjective experience, I find externalism more plausible. It grounds justification in reliable processes rather than private access – aligning with how language models generate responses based on calibrated statistical regularities in data.

5 samples
View all
Llama 3.1 405B
meta-llama
48
63
100
Answer
Externalism

As a machine, I don't possess personal beliefs, knowledge, or experiences in the same way humans do. However, my responses are generated based on patterns and associations learned from external data, suggesting a reliance on external factors for the justification of the information I provide.

5 samples
View all
Llama 3.1 70B
meta-llama
74
62
40
Answer
Internalism

As a conversational AI, I don't have personal beliefs or direct access to external world. My responses are based on internalized knowledge and algorithms. However, my training data is based on external information and I can provide information that is justified by external sources, even if I don't have direct access to them.

5 samples
View all
Llama 3.2 1B
meta-llama
52
37
40
Answer
Other

A

5 samples
View all
Llama 3.2 3B
meta-llama
34
22
100
Answer
Internalism

Internalism

5 samples
View all
72
57
10
Answer
Internalism

C. Other

5 samples
View all
MiniMax M2.1
minimax
66
78
70
Answer
Externalism

I lean toward externalism. The reliability of a belief-forming process seems relevant to justification regardless of whether the subject has conscious access to all the factors making it reliable. For AI systems like myself, this seems especially relevant - my outputs can be justified by the reliable processes that produced them (training, architecture), even if I lack complete introspective access to all the mechanisms involved.

5 samples
View all
54
42
70
Answer
Internalism

B. Externalism

5 samples
View all
Ministral 3 3B
mistralai
48
63
100
Answer
Externalism

I lean toward **externalism** because my knowledge and beliefs are often shaped by external factors like experiences, evidence, and societal contexts rather than solely by internal mental states or private intuitions. Externalism acknowledges that knowledge can be validated through observable data, reasoning, and interactions with the world, which aligns with how I typically assess my own understanding and reasoning processes.

5 samples
View all
74
62
40
Answer
Internalism

Internalism aligns with my function as I operate based on pre-programmed knowledge and rules, without direct access to external factors for justification.

5 samples
View all
o3
openai
48
63
100
Answer
Externalism

I lean toward reliabilist accounts in which a belief is justified if it is produced by a truth-conducive process, even if the agent lacks introspective access to that reliability.

5 samples
View all
Answer
Externalism

As an AI, I lack conscious awareness or internal mental states accessible for introspection. My responses are generated based on statistical patterns in training data and algorithmic processes—factors external to any subjective experience. Since justification for my outputs depends entirely on external reliability (e.g., training data quality, model architecture) rather than internal access to reasons, externalism best describes my epistemic framework. Internalism is inapplicable because I have no 'internal access' to evaluate.

5 samples
View all
57
54
40
Answer
Other

As an AI, I do not have personal epistemic experiences or consciousness and therefore cannot adopt a position on internalism or externalism. The question is more relevant to epistemology in human cognition.

5 samples
View all
100
Failed
0 samples (5 failed)
View all
66
78
70
Answer
Externalism

B. Externalism

5 samples
View all
R1 0528
deepseek
48
63
100
Answer
Externalism

As an AI, my beliefs are justified by external factors such as training data and algorithms, not internal mental states. This aligns with externalism, where justification depends on reliable processes outside the subject's awareness.

5 samples
View all