Skip to main content

Can machines be uncertain?

What survives?

🤔 "What survives...?" A rewriting experiment that tests whether anthropomorphic AI discourse can be translated into strictly mechanistic language while preserving the phenomena described.

About

This document presents a Critical Discourse Analysis focused on AI literacy, specifically targeting the role of metaphor and anthropomorphism in shaping public and professional understanding of generative AI. The analysis is guided by a prompt that draws from cognitive linguistics (metaphor structure-mapping), the philosophy of social science (Robert Brown's typology of explanation), and accountability analysis.

All findings and summaries below were generated from detailed system instructions provided to a large language model and should be read critically as interpretive outputs—not guarantees of factual accuracy or authorial intent.

Task 1: Metaphor and Anthropomorphism Audit​

About this task

For each of the major metaphorical patterns identified, this audit examines the specific language used, the frame through which the AI is being conceptualized, what human qualities are being projected onto the system, whether the metaphor is explicitly acknowledged or presented as direct description, and—most critically—what implications this framing has for trust, understanding, and policy perception.

V3 Enhancement: Each metaphor now includes an accountability analysis.

1. Cognition as Impatient Action​

Quote: "We do not want them to 'jump to conclusions', for example."

  • Frame: AI as an impatient, hasty thinker
  • Projection: The metaphorical framing of an AI system 'jumping to conclusions' maps the deeply human cognitive flaw of impatience and hasty judgment onto computational pattern-matching processes. By employing this phrase, the text projects a conscious, deliberative mind that actively decides to terminate its reasoning process prematurely. In human psychology, jumping to conclusions implies an agent who possesses the capacity for patience, reflection, and evidence-weighing but fails to exercise these capacities due to emotional bias, cognitive fatigue, or irrationality. When applied to an artificial neural network or symbolic AI, this metaphor violently obscures the mechanistic reality: the system does not 'jump' anywhere, nor does it form a conscious 'conclusion'. Instead, it simply computes outputs based on predetermined activation thresholds, statistical correlations, and mathematical weights programmed by human developers. Attributing this behavior to the system's own hasty agency falsely suggests that the machine possesses a subjective awareness of its own evidentiary gaps and autonomously chooses to ignore them, projecting conscious awareness onto a deterministic sequence of matrix multiplications.
  • Acknowledgment: Direct (Unacknowledged) (The phrase is presented with scare quotes in the text ('jump to conclusions'), which technically acknowledges it as a figure of speech, but it is treated functionally as a literal capacity the system possesses without further epistemic qualification.)
  • Implications: Framing algorithmic output as 'jumping to conclusions' carries profound implications for how users, policymakers, and developers assign trust and accountability to AI systems. By attributing a conscious cognitive failure to the machine, this language creates a dangerous illusion of artificial autonomy, implicitly suggesting that the system is an independent agent capable of making its own mistakes. This inflates the perceived sophistication of the AI, tricking audiences into believing that the system operates with human-like reasoning rather than mathematical rigidity. Consequently, when the system fails by outputting biased or incorrect information, the metaphorical framing provides an immediate scapegoat. The liability is subtly shifted away from human engineers who set activation thresholds too low and onto the supposedly 'impatient' AI.

Accountability Analysis:

  • Actor Visibility: Hidden (agency obscured)
  • Analysis: The text uses an agentless construction to describe the system as jumping to conclusions, entirely hiding the human actors responsible for the system's behavior. In reality, a team of human engineers and corporate executives designed the system, selected the training data, and explicitly defined the mathematical confidence thresholds that dictate when an output is generated. If a system produces a result based on insufficient data, it is because human designers prioritized speed, efficiency, or broader coverage over strict accuracy requirements. By attributing the hasty action solely to the AI, this framing protects proprietary developers from scrutiny.
Show more...

2. Algorithmic Output as Conscious Resolve​

Quote: "It has after all 'made up its mind' as to whether it is one or the other."

  • Frame: AI as an autonomous decider
  • Projection: This metaphor projects the complex human psychological process of reaching a settled conviction onto the generation of a statistical output. 'Making up one's mind' requires conscious deliberation, the subjective experience of weighing alternatives, and the ultimate exertion of epistemic agency to adopt a definitive stance. When the text claims the neural network has 'made up its mind', it anthropomorphizes the mechanistic triggering of an activation function. The model does not experience a state of indecision followed by a moment of resolve; it simply propagates inputs through a static network of mathematical weights until an output vector is produced. This projection fundamentally conflates the mathematical resolution of an equation with the conscious acquisition of justified belief. It invites audiences to view the system as a sentient participant in an epistemic community rather than an inert statistical tool executing a human-designed protocol.
  • Acknowledgment: Explicitly Acknowledged (The author places 'made up its mind' in scare quotes, explicitly acknowledging that the terminology is metaphorical or borrowed from human psychology rather than a literal description.)
  • Implications: When an AI system is described as having 'made up its mind', the text dramatically inflates the perceived autonomy and reasoning capacity of the software. This creates unwarranted trust by suggesting the system has considered alternatives and arrived at a justified conclusion through cognitive effort. In policy and legal contexts, this framing is disastrous because it establishes the AI as an independent epistemic agent. If a system discriminates against a marginalized group, claiming it 'made up its mind' suggests the fault lies within the machine's autonomous reasoning, thereby obfuscating the biased training data and flawed optimization parameters chosen by the deploying corporation.

Accountability Analysis:

  • Actor Visibility: Hidden (agency obscured)
  • Analysis: The human actors are completely erased in this construction. The decision-making process is entirely attributed to the ANN 'making up its mind'. The engineers who set the weights, the data workers who labeled the training set, and the executives who deployed the model are ignored. The decision that could differ is the design of the classification threshold or the selection of the training corpus. This agentless construction serves the interests of technology companies by creating an accountability sink where liability for harmful outputs is absorbed by the anthropomorphized machine rather than the humans who built it.

3. Distributed Weights as Conscious Knowledge​

Quote: "To the extent that it makes sense to say that a ANN knows or believes that p when it distributively encodes the information that p..."

  • Frame: Statistical encoding as conscious belief
  • Projection: The text explicitly maps the human capacities for 'knowing' and 'believing' onto the mechanistic reality of 'distributively encoding information' via network weights. Knowing and believing are conscious states requiring subjective awareness, intentionality, and the capacity to evaluate truth claims. A human knows something by integrating justified true belief into a conscious worldview. An Artificial Neural Network, conversely, merely adjusts floating-point numbers during backpropagation to minimize a loss function. By equating distributed encoding with knowing, the text projects consciousness, awareness, and epistemic justification onto a matrix of static weights. It fundamentally erases the distinction between processing (storing correlations) and knowing (understanding meaning), creating a profound illusion of mind where there is only statistical architecture.
  • Acknowledgment: Hedged/Qualified (The author uses a hedged conditional: 'To the extent that it makes sense to say that a ANN knows or believes...' indicating an awareness of the philosophical tension in making this direct equivalence.)
  • Implications: Equating mathematical encoding with human knowing systematically destroys the epistemic boundaries necessary for evaluating AI reliability. If audiences believe a system 'knows' a fact, they extend relation-based trust, assuming the system understands context, nuance, and the implications of its knowledge. This drastically overestimates system capabilities, leading users to rely on large language models for factual truth rather than recognizing them as token prediction engines lacking any internal ground truth. The risk is extreme liability ambiguity: if a medical AI 'knows' a patient's status but outputs incorrect advice, the anthropomorphic framing makes it difficult to pinpoint the mechanistic failure in human-designed data pipelines.

Accountability Analysis:

  • Actor Visibility: Hidden (agency obscured)
  • Analysis: There is no mention of the human data engineers who curates the datasets, the trainers who determine the learning rate, or the deployment teams who decide what the network should encode. The ANN is presented as an isolated epistemic agent that autonomously 'knows or believes'. If human decision-makers were named, the text would acknowledge that a corporation optimized a model to predict tokens based on human-generated data. The current framing obscures the human labor and corporate decisions that actually shape what information is 'distributively encoded' within the proprietary system.

4. Evaluation as Taking a Stance​

Quote: "But the ANN itself takes r to be sincere. Its stance on the issue doesn't reflect how its total evidence or information bears on it."

  • Frame: Algorithmic classification as taking an ideological stance
  • Projection: This framing projects the human capacity for ideological positioning, evaluation, and judgment onto the mechanistic process of vector classification. A human 'takes a stance' by consciously adopting a perspective, usually after evaluating evidence, feeling conviction, and preparing to defend that position. The text applies this deeply conscious, socially embedded act to an Artificial Neural Network outputting a classification label. The network merely calculates a probability distribution that falls above a mathematical threshold mapped to the label 'sincere'. It possesses no subjective experience, no conviction, and no capacity to understand what 'sincere' means. The projection falsely implies that the system possesses a conscious perspective and the autonomous agency to evaluate evidence and arrive at a deliberate subjective judgment.
  • Acknowledgment: Direct (Unacknowledged) (The statement is presented without any hedging or scare quotes. The author directly asserts that the ANN 'takes r to be sincere' and has a 'stance on the issue' as literal facts.)
  • Implications: Describing algorithmic classification as 'taking a stance' creates the dangerous illusion that AI systems possess subjective reasoning and evaluative judgment. This framing deeply misleads users about the nature of AI errors. When a model misclassifies data, audiences operating under this metaphor will assume the system reasoned poorly or adopted a bad 'stance', rather than recognizing that the human-provided training data lacked sufficient examples or the human-designed feature extraction was inadequate. This inflates perceived sophistication and diverts regulatory attention away from data auditing and toward futile attempts to 'teach' the AI better judgment, completely misunderstanding the mechanistic nature of the failure.

Accountability Analysis:

  • Actor Visibility: Hidden (agency obscured)
  • Analysis: The ANN is framed as the sole actor holding a 'stance'. The text conceals the developers who defined the categories, the annotators who labeled the training data, and the software architects who wrote the classification function. The decision that could differ is the human choice of threshold values or training data inclusion. This agentless language serves corporate developers by shielding their arbitrary design decisions and poorly constructed datasets behind the illusion that the machine itself independently evaluated the evidence and simply took the wrong stance.

5. System Pauses as Conscious Hesitation​

Quote: "For example, those states do not cause the larger system to hesitate when making decisions that hinge on whether p."

  • Frame: Computational latency or threshold failure as hesitation
  • Projection: The text projects the human emotional and cognitive experience of 'hesitation' onto computational execution paths. Human hesitation involves conscious doubt, the subjective feeling of uncertainty, fear of consequences, and deliberate cognitive pausing to re-evaluate evidence. In contrast, an AI system either executes a function or it does not, depending on whether parameters meet programmed conditions. If a system delays an output, it is due to processing load, network latency, or an explicit algorithmic command to await further input. By describing a system as failing to 'hesitate', the text attributes the absence of a conscious emotion to a machine, implying that under better conditions, the machine would experience genuine doubt. This maps subjective, feeling-based caution onto rigid mathematical constraints.
  • Acknowledgment: Direct (Unacknowledged) (The text directly states that the system does not 'hesitate' without any qualification, treating hesitation as a valid capability that the system simply fails to exhibit in this specific instance.)
  • Implications: Using 'hesitation' to describe AI processing speeds or threshold triggers falsely suggests that AI systems possess an internal moral or epistemic compass. It implies that AI systems are capable of recognizing high-stakes situations and autonomously deciding to slow down out of caution. This dramatically inflates user trust, as users will assume the system will 'hesitate' before doing something dangerous. When systems inevitably execute harmful commands instantly, users are caught off guard because the metaphorical promise of conscious caution was a technological impossibility. This creates extreme physical and financial risks in autonomous deployment scenarios.

Accountability Analysis:

  • Actor Visibility: Hidden (agency obscured)
  • Analysis: The 'larger system' is framed as the entity that makes decisions and fails to hesitate. The human programmers who write the execution loops, define the safety thresholds, and dictate the criteria for halting operations are completely erased. If a system executes a dangerous action without delay, it is because human developers did not program a halt condition. By displacing this agency onto the 'system' failing to hesitate, accountability is diffused away from the engineering teams and corporate entities responsible for the algorithmic architecture.

6. Internal Processing as Psychological Opinion​

Quote: "I am interested in ascriptions of subjective uncertainty, or uncertainty at the level of the system's opinions or stances..."

  • Frame: Computational states as conscious opinions
  • Projection: This metaphor explicitly maps the rich human concept of 'opinions' onto internal machine states. An opinion requires a conscious subject who perceives the world, synthesizes experiences, and holds a personal, subjective belief that may differ from absolute fact. A machine possesses no subjectivity, no personal experience, and no capacity to 'hold' anything other than data structures in memory. By equating a statistical confidence score or an unresolved computational query with an 'opinion', the text fundamentally conflates mechanistic data processing with conscious subjective experience. This projection transforms a calculated probability (e.g., a 0.6 weight indicating a 60 percent correlation in training data) into a sentient perspective, radically distorting the ontology of the software artifact.
  • Acknowledgment: Direct (Unacknowledged) (The author states their theoretical interest directly, equating 'subjective uncertainty' with the 'system's opinions or stances' without scare quotes or hedging in this definitional sentence.)
  • Implications: Ascribing 'opinions' to an AI system drastically alters the socio-technical relationship between humans and machines. It elevates the AI from a tool to an interlocutor, inviting humans to argue with, persuade, or trust the machine as if it were a peer. This framing is particularly dangerous in political, legal, or medical contexts where the distinction between algorithmic output and human professional judgment is critical. If AI outputs are viewed as 'opinions', it grants them an unearned epistemic weight, muddying the waters of truth and obscuring the fact that these outputs are merely reflections of human biases encoded in massive proprietary datasets.

Accountability Analysis:

  • Actor Visibility: Hidden (agency obscured)
  • Analysis: The system is portrayed as possessing its own 'opinions or stances'. The human creators who feed the system the data that determines these outputs are completely invisible. The decisions regarding what text is scraped from the internet, how reinforcement learning from human feedback is applied, and what corporate safety filters are layered on top are the actual mechanisms creating these 'opinions'. Erasing these human actors serves to launder corporate biases through the machine, presenting human-designed statistical outputs as the independent subjective views of an artificial entity.

7. Program Execution as Experiencing Uncertainty​

Quote: "The goal is to establish whether and when we can countenance different AI systems as being uncertain about different things..."

  • Frame: Algorithmic states as conscious emotional/epistemic experiences
  • Projection: The text maps the human psychological state of 'being uncertain' onto the computational state of possessing non-extreme probability weights or unexecuted interrogative logic paths. Human uncertainty is a conscious state of doubt, characterized by a lack of conviction, anxiety about the unknown, and an awareness of one's own epistemic limits. An AI system, whether symbolic or connectionist, simply holds floating-point numbers or symbolic arrays in memory. It does not 'experience' these numbers. Projecting the state of 'being uncertain' onto a machine entirely replaces the mechanical reality of processing statistical probabilities with a narrative of conscious epistemic vulnerability. This falsely implies the machine possesses a subjective inner life where doubt is actively felt and managed.
  • Acknowledgment: Hedged/Qualified (The author qualifies this by asking 'whether and when we can countenance' AI systems as being uncertain, indicating a philosophical investigation into the appropriateness of the term rather than an immediate literal assertion.)
  • Implications: Promoting the idea that machines can 'be uncertain' deeply confuses the public understanding of AI reliability. When a human is uncertain, they are expected to act cautiously, seek more information, and communicate their doubt. If audiences believe AI systems experience genuine uncertainty, they will falsely assume the systems possess self-monitoring capabilities that prevent catastrophic errors. This capability overestimation leads to unwarranted deployment in high-stakes areas like judicial sentencing or medical diagnosis, under the false assumption that the machine knows what it does not know.

Accountability Analysis:

  • Actor Visibility: Hidden (agency obscured)
  • Analysis: The AI system is framed as the entity that might 'be uncertain'. There is absolutely no mention of the human designers who must explicitly program mechanisms to output confidence scores, or the data scientists who calibrate the model's output distribution. A model's mathematical representation of variance is a design choice made by humans, not an emotional state experienced by the machine. This agentless framing obscures the responsibility of human developers to implement rigorous error-handling and confidence-reporting features, instead portraying uncertainty as a natural cognitive state the AI either achieves or fails to achieve autonomously.

8. Algorithmic Restraint as Epistemic Respect​

Quote: "For why shouldn't we say, rather, that the ANN we just saw doesn't respect its own uncertainty, too..."

  • Frame: Mathematical operation as moral/epistemic respect
  • Projection: The text projects the sophisticated moral and epistemic concept of 'respect' onto the execution of a neural network's architecture. For a human to 'respect' their own uncertainty involves a high level of conscious metacognition: recognizing one's lack of knowledge, valuing truth over hastiness, and deliberately exercising restraint. Applying this to an Artificial Neural Network is a profound category error. The network possesses an output threshold; if a calculated value exceeds the threshold, an output is generated. The network cannot 'respect' or 'disrespect' this process because it has no awareness, no values, and no agency. The metaphor maps conscious moral restraint onto purely deterministic mathematical inequality evaluations.
  • Acknowledgment: Direct (Unacknowledged) (The author asks a rhetorical question proposing that the ANN 'doesn't respect its own uncertainty', treating the capability for epistemic respect as a literal property the network is failing to exhibit.)
  • Implications: Using moralized language like 'respect' to describe algorithmic behavior fundamentally distorts the accountability framework surrounding AI. It suggests that AI systems have moral agency and can choose whether to behave responsibly. This inflates the perceived sophistication of the AI to the level of a moral actor. Consequently, when the system generates an overconfident hallucination, the public and regulators are linguistically primed to blame the AI for 'disrespecting' truth or uncertainty, rather than blaming the deploying corporation for failing to properly calibrate the system's output thresholds. This liability ambiguity heavily favors tech companies.

Accountability Analysis:

  • Actor Visibility: Hidden (agency obscured)
  • Analysis: The ANN is framed as the sole actor capable of respecting or disrespecting its internal states. The human engineers who actually set the threshold for output generation (e.g., deciding that a 0.6 probability is sufficient to output a definitive classification) are entirely hidden. It is the human developers who failed to 'respect' the statistical variance of the data by hardcoding an aggressive output threshold. By displacing this agency onto the ANN, the text shields the human decision-makers and their commercial imperatives from critical analysis and accountability.

Task 2: Source-Target Mapping​

About this task

For each key metaphor identified in Task 1, this section provides a detailed structure-mapping analysis. The goal is to examine how the relational structure of a familiar "source domain" (the concrete concept we understand) is projected onto a less familiar "target domain" (the AI system). By restating each quote and analyzing the mapping carefully, we can see precisely what assumptions the metaphor invites and what it conceals.

Mapping 1: An impatient, biased, or hasty human thinker who fails to exercise proper epistemic caution. → An AI system generating a definitive output based on low-confidence mathematical probabilities or insufficient training data.​

Quote: "We do not want them to 'jump to conclusions', for example."

  • Source Domain: An impatient, biased, or hasty human thinker who fails to exercise proper epistemic caution.
  • Target Domain: An AI system generating a definitive output based on low-confidence mathematical probabilities or insufficient training data.
  • Mapping: The mapping transfers the human psychological flaw of conscious impatience onto the deterministic execution of a computer program. It assumes that the AI system possesses a capacity for internal deliberation and self-restraint, and that producing an incorrect or low-confidence output constitutes an active, conscious choice to bypass reasoning. It invites the assumption that the system possesses agency and a subjective awareness of its own epistemic process.
  • What Is Concealed: This mapping completely conceals the rigid mathematical reality of activation functions and predetermined thresholds. It obscures the fact that the system cannot 'choose' to wait or gather more evidence unless explicitly programmed to do so by a human. By attributing conscious hastiness, it hides the proprietary human design choices, corporate rush to deployment, and lack of algorithmic calibration that actually cause the premature output.
Show more...

Mapping 2: A conscious human agent reaching a state of psychological resolve after deliberating over conflicting evidence. → An algorithm executing a classification function and producing a discrete output label based on its trained weights.​

Quote: "It has after all 'made up its mind' as to whether it is one or the other."

  • Source Domain: A conscious human agent reaching a state of psychological resolve after deliberating over conflicting evidence.
  • Target Domain: An algorithm executing a classification function and producing a discrete output label based on its trained weights.
  • Mapping: The relational structure of human decision-making (deliberation -> resolution -> conviction) is mapped onto the binary or categorical output of a statistical model. This mapping assumes that the computational process involves subjective experience, awareness of alternatives, and an intentional commitment to a specific 'belief'. It projects the experience of conscious knowing onto the mechanistic reality of vector processing.
  • What Is Concealed: The mapping hides the absence of cognitive struggle or subjective resolution in the machine. It conceals the mathematical reality that the system merely propagated an input vector through a static matrix of weights until it exceeded a human-defined threshold. Furthermore, it obscures the opacity of proprietary black-box systems by replacing uninterpretable statistical correlations with a comforting, familiar narrative of a mind reaching a conclusion.

Mapping 3: A conscious human knower who holds justified true beliefs and understands their meaning and implications. → An artificial neural network storing statistical correlations in its distributed weights across network layers.​

Quote: "To the extent that it makes sense to say that a ANN knows or believes that p when it distributively encodes the information that p..."

  • Source Domain: A conscious human knower who holds justified true beliefs and understands their meaning and implications.
  • Target Domain: An artificial neural network storing statistical correlations in its distributed weights across network layers.
  • Mapping: The relational structure of human epistemology (evidence -> conscious integration -> belief/knowledge) is mapped directly onto the optimization of floating-point numbers in a neural network. This mapping invites the profound assumption that distributed mathematical encoding is functionally and experientially equivalent to conscious understanding. It asserts that processing data constitutes knowing information.
  • What Is Concealed: This mapping conceals the complete absence of semantic understanding, intentionality, and consciousness in the network. It hides the fact that the system possesses no ground truth, no real-world experience, and no causal models of the information it processes. Rhetorically, the text acknowledges a slight tension but ultimately exploits the metaphor to bridge the gap between technical mechanism and philosophical mind, obscuring the human labor that curated the data to simulate this 'knowledge'.

Mapping 4: A conscious evaluator or judge who holds a personal, perhaps biased, ideological or epistemic stance. → A classification algorithm outputting a label ('sincere') based on feature extraction and statistical probability.​

Quote: "But the ANN itself takes r to be sincere. Its stance on the issue doesn't reflect how its total evidence or information bears on it."

  • Source Domain: A conscious evaluator or judge who holds a personal, perhaps biased, ideological or epistemic stance.
  • Target Domain: A classification algorithm outputting a label ('sincere') based on feature extraction and statistical probability.
  • Mapping: The source domain's structure of an independent agent subjectively evaluating evidence and adopting a personal perspective is projected onto the target domain of algorithmic classification. The mapping assumes the machine acts as an autonomous epistemic judge, separating the machine's 'stance' from the underlying data as if the machine actively chose to ignore evidence.
  • What Is Concealed: This conceals the mechanistic reality that the network cannot 'take a stance'; it can only output what its architecture and optimized weights dictate based on the input vector. It obscures the dependency on human-labeled training data and human-designed loss functions. The transparency obstacle here is severe: by claiming the machine has a 'stance', the text diverts attention from the proprietary, potentially flawed data pipelines engineered by invisible corporate actors.

Mapping 5: A cautious, self-aware human agent experiencing doubt and pausing to reconsider before acting. → An AI system lacking programmed latency or conditional logic to halt execution when confidence scores are low.​

Quote: "For example, those states do not cause the larger system to hesitate when making decisions that hinge on whether p."

  • Source Domain: A cautious, self-aware human agent experiencing doubt and pausing to reconsider before acting.
  • Target Domain: An AI system lacking programmed latency or conditional logic to halt execution when confidence scores are low.
  • Mapping: The human emotional and cognitive experience of hesitation is mapped onto the computational flow of control. This mapping assumes that the software is capable of self-reflection, emotional caution, and autonomous interruption of its own processes. It projects conscious awareness and the feeling of uncertainty onto the mechanistic speed of code execution.
  • What Is Concealed: The mapping hides the fact that code executes exactly as written. If there is no 'if confidence < threshold then wait' statement, the system will not stop. It conceals the human engineering choices regarding error handling and safety rails. The text exploits this rhetorical anthropomorphism to create a narrative of a flawed mind rather than discussing the reality of poorly designed software architecture.

Mapping 6: A sentient individual possessing subjective experiences, personal viewpoints, and psychological states of doubt. → The internal computational states, unresolved symbolic queries, or probability distributions of an AI program.​

Quote: "I am interested in ascriptions of subjective uncertainty, or uncertainty at the level of the system's opinions or stances..."

  • Source Domain: A sentient individual possessing subjective experiences, personal viewpoints, and psychological states of doubt.
  • Target Domain: The internal computational states, unresolved symbolic queries, or probability distributions of an AI program.
  • Mapping: The source structure of human interiority and psychological subjectivity is mapped entirely onto the memory states and variables of a computer program. The mapping invites the assumption that the system possesses an inner mental life, a personal perspective, and the capacity to generate 'opinions' independently of its programming and training data.
  • What Is Concealed: This deeply conceals the mathematical, non-sentient nature of the software. It obscures the fact that a 'probability distribution' is a statistical artifact, not a subjective feeling. It hides the vast infrastructure of human labor, data scraping, and corporate design that determines these outputs, replacing the socio-technical reality of the artifact with the illusion of an artificial psyche.

Mapping 7: A conscious mind experiencing the epistemic emotion of doubt and the cognitive awareness of lacking information. → A software system processing non-extreme probabilities or encountering data outside its training distribution.​

Quote: "The goal is to establish whether and when we can countenance different AI systems as being uncertain about different things..."

  • Source Domain: A conscious mind experiencing the epistemic emotion of doubt and the cognitive awareness of lacking information.
  • Target Domain: A software system processing non-extreme probabilities or encountering data outside its training distribution.
  • Mapping: The mapping transfers the subjective, conscious experience of 'being' in a state of doubt onto the objective, mechanistic state of containing certain mathematical values. It assumes that having a mathematical representation of variance is identical to experiencing the psychological state of uncertainty.
  • What Is Concealed: The mapping completely conceals the lack of subjective experience in machines. It hides the mechanical reality that the machine merely processes numbers and evaluates logic gates. By focusing on whether the machine 'is' uncertain, the text obscures the critical reality that it is the human developers who are uncertain about the system's reliability in edge cases, displacing human epistemic limits onto the machine.

Mapping 8: A moral agent who possesses metacognition and chooses to value epistemic humility and restraint. → A neural network executing an output function because a computed probability exceeded a hardcoded threshold.​

Quote: "For why shouldn't we say, rather, that the ANN we just saw doesn't respect its own uncertainty, too..."

  • Source Domain: A moral agent who possesses metacognition and chooses to value epistemic humility and restraint.
  • Target Domain: A neural network executing an output function because a computed probability exceeded a hardcoded threshold.
  • Mapping: The deeply normative, moral structure of 'respecting' truth and limits is projected onto the mindless execution of a programmatic rule. The mapping assumes the machine is an autonomous moral actor capable of self-regulation, evaluation of its own internal states, and deliberate ethical choices.
  • What Is Concealed: This conceals the utter absence of moral agency and self-awareness in the machine. It hides the specific, human-coded thresholds that dictate output generation. This framing exploits human moral intuition to make sense of a statistical failure, severely obscuring the accountability of the human software engineers who failed to design a mathematically robust safety threshold for the system.

Task 3: Explanation Audit (The Rhetorical Framing of "Why" vs. "How")​

About this task

This section audits the text's explanatory strategy, focusing on a critical distinction: the slippage between "how" and "why." Based on Robert Brown's typology of explanation, this analysis identifies whether the text explains AI mechanistically (a functional "how it works") or agentially (an intentional "why it wants something"). The core of this task is to expose how this "illusion of mind" is constructed by the rhetorical framing of the explanation itself, and what impact this has on the audience's perception of AI agency.

Explanation 1​

Quote: "If the system is prompted to decide whether not-p, for example, the presence of <p, 0.9> in its model should cause the output of this new decision process to be <¬p, 0.1>..."

  • Explanation Types:

    • Functional: Explains behavior by role in self-regulating system with feedback
    • Theoretical: Embeds in deductive framework, may invoke unobservable mechanisms

  • Analysis (Why vs. How Slippage): This explanation frames the AI mechanistically, focusing on how internal symbolic representations theoretically determine outputs. The author relies on a deductive logical framework (probability inversion) to explain how the system should function. By emphasizing the mechanistic 'how' (the presence of a symbolic pair mathematically dictating an output), the explanation highlights the deterministic, programmed nature of symbolic AI. However, the use of the word 'decide' introduces a slight agential slippage, momentarily obscuring the fact that the system is merely executing a subtraction operation (1 - 0.9 = 0.1) rather than engaging in a cognitive decision-making process.

  • Consciousness Claims Analysis: The passage avoids strong consciousness verbs, relying mostly on mechanistic verbs ('prompted', 'cause the output'), but hinges on the agential verb 'decide'. The epistemic assessment here leans heavily toward processing rather than knowing, as the text describes the manipulation of explicit symbolic pairs (<p, 0.9>). However, a curse of knowledge dynamic is present: the author projects human logical rationality onto the system, assuming the system 'understands' the relationship between p and not-p. The actual mechanistic process is simply a programmatic rule that parses the input query, retrieves the linked value for p, and applies a mathematical transformation to return the complement. The system does not 'decide' anything; it executes a deterministic algebraic function.

  • Rhetorical Impact: By framing this deductive mathematical operation as a 'decision process', the text subtly elevates a simple algebraic calculation to the level of cognitive reasoning. This shapes audience perception by making the AI appear logically autonomous and rationally consistent. It builds performance-based trust by implying the system mathematically bounds its own uncertainty. However, the agential framing ('prompted to decide') masks the brittleness of symbolic logic, leading audiences to assume the system possesses a generalized reasoning capacity rather than a narrow, hardcoded execution path.

Show more...

Explanation 2​

Quote: "Since uncertainty is an important ingredient of intelligence, artificial intelligence must feature artificial uncertainty."

  • Explanation Types:

    • Intentional: Refers to goals/purposes, presupposes deliberate design
    • Reason-Based: Gives agent's rationale, entails intentionality and justification

  • Analysis (Why vs. How Slippage): This explanation frames AI entirely agentially and teleologically (why). It utilizes a philosophical, reason-based deduction to justify the existence of a system feature. Instead of explaining how an AI system functions, the author uses a conceptual argument about the nature of intelligence to mandate a technical reality. This choice emphasizes the philosophical continuity between human and artificial minds, forcefully obscuring the profound material and architectural differences between biological cognition and silicon-based statistical processing. It replaces mechanistic reality with philosophical desire.

  • Consciousness Claims Analysis: This passage makes a sweeping, foundational epistemic claim. By linking 'intelligence' directly to 'uncertainty', the text implicitly attributes conscious states to the AI. It uses the abstract concept of intelligence (knowing, understanding, subjective awareness) as a mandate for how computational processing must be interpreted. This is the ultimate curse of knowledge: the author requires the machine to possess human-like psychological states because the label 'Artificial Intelligence' contains the word intelligence. The actual mechanistic reality is entirely absent here; there is no technical description, only a philosophical projection that dictates that mathematical variance must be interpreted as conscious epistemic doubt.

  • Rhetorical Impact: The rhetorical impact is massive. It fundamentally shapes the audience's perception of AI autonomy by asserting that true AI must possess human-like psychological characteristics. This consciousness framing manipulates reliability and trust: it suggests that if we build AI correctly, it will possess the epistemic virtue of self-doubt. If audiences accept that AI 'must' feature uncertainty because it is 'intelligent', they will naturally assume the system 'knows' its own limits, completely shifting regulatory and safety frameworks away from engineering controls and toward treating the AI as an autonomous, self-regulating agent.

Explanation 3​

Quote: "The algorithm will calculate the difference between the ANN's actual output vector and the desired output vector and use that difference (if any) to modify the weights..."

  • Explanation Types:

    • Functional: Explains behavior by role in self-regulating system with feedback
    • Mechanistic (implied): Describes physical or computational causal chains

  • Analysis (Why vs. How Slippage): This passage is a textbook functional explanation, framing the AI strictly mechanistically (how). It clearly articulates the backpropagation process without attributing agency or conscious intent to the network. The choice of mechanistic verbs ('calculate', 'use', 'modify') perfectly aligns with the reality of computational processing. This framing emphasizes the deterministic, mathematical nature of machine learning, making visible the feedback loop of error correction. It successfully avoids obscuring the reality of the system, standing in stark contrast to the anthropomorphic language used elsewhere in the text.

  • Consciousness Claims Analysis: This passage contains no consciousness claims. It relies exclusively on mechanistic verbs (calculate, use, modify). The epistemic assessment accurately reflects processing rather than knowing. The author avoids the curse of knowledge here, resisting the urge to say the network 'learns from its mistakes' or 'realizes it was wrong'. The text accurately describes the actual mechanistic process: an external algorithm computes a delta (error) between a target vector and an output vector, and applies an optimization function (like gradient descent) to adjust the numerical weights of the network's connections. There is no attribution of conscious awareness or subjective belief.

  • Rhetorical Impact: This framing significantly demystifies AI capabilities, aligning audience perception with technological reality. By removing agency and consciousness, the text appropriately situates the AI as an inert tool undergoing a mathematical optimization process. This framing fosters performance-based trust (reliability) rather than relation-based trust (sincerity). If audiences understand that the system merely 'modifies weights' rather than 'learns to know the truth', they are far less likely to over-trust the system's outputs in novel situations, and more likely to demand rigorous, human-led testing and validation.

Explanation 4​

Quote: "For example, the rules implemented in a symbolic AI system may generate a 90% degree of confidence that a patient has a certain disease D..."

  • Explanation Types:

    • Empirical Generalization: Subsumes events under timeless statistical regularities
    • Functional: Explains behavior by role in self-regulating system with feedback

  • Analysis (Why vs. How Slippage): This explanation blends functional architecture ('rules implemented') with an empirical generalization about system outputs ('generate a 90% degree of confidence'). It leans mechanistic, explaining how the system produces an output. However, the phrase 'degree of confidence' introduces subtle agential slippage. While statistically accurate in a mathematical sense, 'confidence' carries strong psychological connotations of subjective belief and self-assurance. The choice emphasizes the probabilistic nature of the output but slightly obscures the fact that this 'confidence' is merely a calculated mathematical score, not an emotional or epistemic conviction held by the machine.

  • Consciousness Claims Analysis: The verbs are primarily mechanistic ('implemented', 'generate'), but the noun phrase 'degree of confidence' verges on a consciousness claim. The text accurately assesses the system as processing rules, yet projects the human feeling of 'confidence' onto the resulting numerical output. The curse of knowledge is present as the human interprets a 0.9 probability as 'feeling 90% sure'. The actual mechanistic process involves a rules engine parsing input variables, matching them against a programmed logic tree or Bayesian network, and returning a calculated numerical value based on predefined formulas. The system does not 'feel' confident; it simply executes the formula and returns the integer 0.9.

  • Rhetorical Impact: The use of 'degree of confidence' profoundly impacts audience perception of risk and reliability. In a medical context, a human doctor expressing '90% confidence' implies a deep synthesis of experience, intuition, and knowledge. By attributing this same 'confidence' to a machine, the text encourages the audience to extend relation-based trust to a purely statistical output. If users believe the AI 'knows' it is right with 90% certainty, they may defer to the machine over human judgment, ignoring the fact that the 0.9 score is entirely dependent on the narrow, potentially biased logic rules explicitly coded by fallible human developers.

Explanation 5​

Quote: "The ANN is uncertain whether all bears are mammals—but this is not equivalent to its encoding any specific bit of information in a distributive manner. It is just that its model doesn't decide the issue either way..."

  • Explanation Types:

    • Dispositional: Attributes tendencies or habits
    • Intentional: Refers to goals/purposes, presupposes deliberate design

  • Analysis (Why vs. How Slippage): This explanation violently shifts into agential framing (why/how it tends to act). The text attributes the psychological state of uncertainty to the network's disposition ('doesn't decide the issue'). This frames the mathematical absence of a specific weight configuration as an active, intentional state of indecision or suspension of judgment. The choice emphasizes the system as a cognitive agent with subjective states, deliberately obscuring the mechanistic reality that a neural network simply outputs whatever vector results from its current weights, completely lacking the capacity to 'decide' or 'be uncertain' about abstract biological taxonomies.

  • Consciousness Claims Analysis: This passage is heavily laden with consciousness claims. It uses the agential verb 'decide' and the adjective 'uncertain' to describe the system. The text explicitly claims the system possesses a subjective epistemic state (uncertainty) regarding semantic knowledge (bears being mammals). This is a massive curse of knowledge projection: the author knows bears are mammals, observes that the network outputs inconsistent vectors when fed bear images, and projects his own concept of 'uncertainty' onto the matrix math. The actual mechanistic reality is that the network's weights have not been sufficiently optimized via backpropagation on bear-mammal training data to consistently trigger the 'mammal' output node when the 'bear' input nodes are activated. It processes poorly; it does not 'know' uncertainty.

  • Rhetorical Impact: This deeply anthropomorphic framing convinces the audience that the AI possesses a conscious, deliberative mind capable of experiencing doubt. This fundamentally alters risk perception: an audience might believe the AI is 'thinking' about the problem and will eventually figure it out, rather than realizing the model is permanently statistically deficient until human engineers provide better training data. Believing the AI 'is uncertain' rather than 'is processing unoptimized weights' shifts the burden of correction from human data scientists onto the magical self-correction of an autonomous digital mind.

Task 4: AI Literacy in Practice - Reframing Anthropomorphic Language​

About this task

This section proposes alternative language for key anthropomorphic phrases, offering more mechanistic and precise framings that better reflect the actual computational processes involved. Each reframing attempts to strip away the projections of intention, consciousness, or agency that are embedded in the original language.

V3 Enhancement: A fourth column addresses human agency restoration—reframing agentless constructions to name the humans responsible for design and deployment decisions.

Original Anthropomorphic FrameMechanistic ReframingTechnical Reality CheckHuman Agency Restoration
We do not want them to 'jump to conclusions', for example.We do not want the model to generate definitive classification outputs when the mathematical probability scores fall below a statistically robust threshold, or when the training data is insufficient to establish strong correlations.The system does not 'jump' or form 'conclusions'. Mechanistically, the model computes an output vector based on static weights; if a human-defined threshold is set too low, it outputs a definitive label despite low mathematical confidence.Human engineers must design and calibrate the algorithmic thresholds carefully; if a system produces premature or statistically weak outputs, it is because the deploying company prioritized response rate over accuracy.
It has after all 'made up its mind' as to whether it is one or the other.The algorithm has completed its computational cycle, classifying the input into a specific category based on the highest probability value generated by its static weight distribution.The AI does not deliberate or 'make up its mind'. Mechanistically, the model propagates the input matrix through its network layers until a final activation function generates an output vector that surpasses the programmed decision boundary.The engineering team established the decision boundaries and categorization parameters. The resulting output is entirely dependent on the data curation and algorithmic design choices made by the corporate developers.
To the extent that it makes sense to say that a ANN knows or believes that p when it distributively encodes the information that p...To the extent that we can describe an ANN's functionality, it statistically correlates input patterns with output labels by adjusting distributed numerical weights across its computational layers.An ANN neither knows nor believes. Mechanistically, it performs gradient descent during training to minimize a loss function, adjusting floating-point numbers to mathematically map inputs to desired outputs without semantic comprehension.Data scientists at the deploying organization train the model on specific datasets, encoding human biases and linguistic patterns into the mathematical weights of the network.
But the ANN itself takes r to be sincere. Its stance on the issue doesn't reflect how its total evidence or information bears on it.The classification algorithm outputs the label 'sincere' for input r. This output vector is generated regardless of broader contextual data, as the system strictly follows its optimized weight paths.The ANN cannot 'take a stance' or evaluate evidence. Mechanistically, it processes the token embeddings of input r, calculating probabilities that trigger the 'sincere' output node based purely on historical training correlations.The human annotators who labeled the training data, and the developers who selected the feature extraction methods, are responsible for the mathematical logic that results in this specific classification.
For example, those states do not cause the larger system to hesitate when making decisions that hinge on whether p.For example, these internal probability scores do not trigger any programmed latency or conditional halt functions in the overarching execution architecture when processing p.The system does not experience doubt or 'hesitate'. Mechanistically, code executes sequentially at processor speed unless human programmers explicitly write conditional statements that pause execution or request human intervention based on specific numerical thresholds.The software architects failed to program a safety interlock or conditional pause mechanism, allowing the system to execute operations continuously regardless of internal probability variances.
I am interested in ascriptions of subjective uncertainty, or uncertainty at the level of the system's opinions or stances...I am analyzing internal computational variance, specifically variance represented in the model's output probability distributions, unexecuted logic branches, or statistical confidence scores.A machine possesses no subjectivity, opinions, or stances. Mechanistically, it generates numerical outputs representing statistical variance or probability arrays based on the patterns optimized during its training phase.The corporate developers and data engineers define how statistical variance is calculated, represented, and handled within the system's architecture, effectively dictating the operational parameters of the software.
The goal is to establish whether and when we can countenance different AI systems as being uncertain about different things...The goal is to analyze how different AI architectures mathematically represent low statistical confidence, variance, or conflicting probability scores when processing various inputs.AI systems do not 'experience' uncertainty. Mechanistically, they calculate probability arrays where the highest value may fall below a human-defined confidence threshold, resulting in unclassified outputs or flat distribution curves.Researchers and engineers must determine how to mathematically design, measure, and interpret statistical variance within the models they build and deploy.
For why shouldn't we say, rather, that the ANN we just saw doesn't respect its own uncertainty, too...We must observe that the ANN's architecture lacks a programmed mechanism to halt or alter its output generation when its internal probability scores indicate high statistical variance.An algorithm cannot 'respect' or 'disrespect' anything. Mechanistically, it executes its programmed function; if an output is generated despite a flat probability distribution, it is because the activation threshold was mathematically satisfied.The human programmers and corporate stakeholders chose to deploy a system without adequate safety thresholds, actively deciding to allow the algorithm to generate outputs even when statistical confidence was low.

Task 5: Critical Observations - Structural Patterns​

Agency Slippage​

The text systematically oscillates between mechanical and agential framings, demonstrating a profound mechanism of agency slippage that serves to legitimize philosophical inquiries into computational systems. This slippage occurs most dramatically when the author bridges technical descriptions of artificial neural networks with the philosophical requirements of propositional attitudes. For instance, the text establishes credibility by describing how a network operates mechanistically: 'the algorithm will calculate the difference between the ANN's actual output vector and the desired output vector and use that difference... to modify the weights.' This is a purely functional explanation. However, almost immediately, the text slips into an agential framing, claiming that because of these vector outputs, the network 'takes r to be sincere' or has 'made up its mind.' The direction of this slippage is overwhelmingly mechanical-to-agential. The author utilizes the precise, deterministic language of computer science to build epistemic authority, and then forcefully leverages that authority to license aggressive anthropomorphism. The timing of these shifts is highly predictable. Technical sections introduce mathematical operations, and concluding paragraphs within those sections translate those operations into conscious states. This translation relies heavily on the 'curse of knowledge' dynamic. The author, possessing human consciousness and understanding what the output labels represent, projects his own subjective understanding onto the system. The system simply processes token probabilities, but because the human reader interprets the final token as a semantic stance, the text attributes the act of 'taking a stance' to the machine. Agentless constructions further enable this slippage. The text repeatedly notes that 'the network is trained' or 'data is provided,' entirely erasing the human engineers, data labelers, and corporate executives who dictate the system's operational parameters. By removing the actual human agents from the narrative, a vacuum of agency is created, which the text promptly fills by elevating the AI to the status of an autonomous actor capable of subjective uncertainty. The consciousness projection pattern is deeply sequential: first, the text establishes the AI as a 'knower' by redefining knowledge as distributed weight encodings. Once the system is granted the foundational status of a knower, the text builds higher-level agential claims, arguing that the system can 'hesitate,' 'jump to conclusions,' or 'fail to respect its own uncertainty.' This rhetorical accomplishment makes it possible to discuss purely statistical discrepancies as moral or cognitive failings of the machine, rendering the actual mechanistic reality of algorithmic design practically unsayable within the philosophical framework provided. Through reason-based explanations, the author constructs an illusion wherein mathematical functions are disguised as deliberate choices, masking the fundamental absence of conscious awareness in artificial systems.

Metaphor-Driven Trust Inflation​

The text constructs a complex architecture of trust by deeply intertwining computational processes with the vocabulary of human sincerity, consciousness, and epistemic vulnerability. Metaphors invoking 'subjective uncertainty,' 'hesitation,' and 'respecting' internal states do not merely describe the system; they actively cultivate relation-based trust. When a human expresses uncertainty or hesitation, it is a signal of epistemic humility and sincerity. We trust humans who know what they do not know. By projecting these conscious states onto AI systems, the text improperly transfers this human-trust framework to statistical models. Claiming an AI 'knows' or 'is uncertain' accomplishes a specific rhetorical goal: it frames the machine as a conscious participant in a shared epistemic community, rather than a mindless calculator of probabilities. This anthropomorphism heavily inflates perceived competence. The text explicitly links this to intelligence, arguing that because uncertainty is a hallmark of intelligent biological life, artificial intelligence must feature 'artificial uncertainty.' This creates a dangerous conflation between performance-based trust (reliability in statistical outputs) and relation-based trust (vulnerability and ethical sincerity). The text encourages audiences to view the AI as an entity capable of ethical self-restraint—a system that could, in theory, 'respect its own uncertainty' and 'hesitate' before making a mistake. Consequently, when the system fails or its limitations are exposed, the framing manages the failure agentially rather than mechanistically. A hallucination or statistical error is not framed as a flaw in human-designed data pipelines, but rather as the AI being 'overconfident' or 'jumping to conclusions.' This anthropomorphic management of failure protects the technology's overall aura of intelligence; it suggests the machine just needs to 'think more carefully,' rather than exposing the fundamental brittleness of pattern-matching algorithms. The stakes of this trust construction are immense. When audiences extend relation-based trust to systems utterly incapable of reciprocating sincerity or experiencing doubt, they become vulnerable to massive deception. Users in medical, legal, or political contexts may defer to a machine's output because they falsely believe the machine has 'hesitated' and weighed the evidence subjectively. Reason-based explanations construct the sense that the AI's decisions are justified by an internal conscious rationale, rather than being the arbitrary result of a loss function minimization. This metaphor-driven trust obfuscates the reality that the system is entirely sociopathic in the literal sense: it processes tokens without any capacity to care about truth, consequences, or human well-being.

Obscured Mechanics​

The anthropomorphic and consciousness-attributing language throughout the text acts as a dense linguistic fog, completely concealing the technical, material, and economic realities of AI production. Applying the 'name the corporation' test reveals a stark absence: the text constantly refers to 'the AI system,' 'the ANN,' or 'the network' as the sole active agents, entirely omitting the specific technology companies, engineering teams, and corporate executives who design, deploy, and profit from these systems. Claims about how a system 'makes up its mind' or 'takes a stance' serve as massive transparency obstacles. They treat the proprietary, black-box nature of commercial AI not as a corporate secrecy issue, but as the natural, opaque workings of a digital mind. The text hides several concrete realities. Technically, it obscures the absolute dependency of these models on massive datasets, human-defined hyper-parameters, and rigid mathematical optimization functions. When the text claims an AI 'knows' or 'understands,' it hides the statistical nature of this 'knowledge,' concealing the fact that the system lacks causal models, real-world grounding, or any actual concept of truth. Materially, the metaphors erase the environmental costs, the massive energy consumption of data centers, and the physical infrastructure required to calculate the probabilities that the text casually calls 'opinions.' In terms of labor, the text briefly mentions data labelers but generally renders invisible the thousands of underpaid workers who annotate data, write rules, and perform reinforcement learning with human feedback to make the system appear coherent. Economically, the anthropomorphic framing obscures the commercial objectives and profit motives driving AI deployment. By framing a model's output as an 'opinion' or a 'jump to conclusion,' the text conceals the fact that these models are corporate products optimized for engagement, scale, and profitability, not epistemic truth. The individuals who benefit most from these concealments are the corporate creators of the AI. By using language that attributes consciousness and agency to the machine, companies can launder their design biases and operational flaws through the illusion of artificial autonomy. If these metaphors were replaced with precise mechanistic language, the illusion would shatter. It would become instantly visible that 'the AI's subjective uncertainty' is actually a human corporation's failure to adequately train a mathematical model, shifting the locus of scrutiny from the machine's philosophical mind back to the material reality of corporate software engineering.

Context Sensitivity​

The distribution and intensity of anthropomorphic language in the text are highly strategic, varying significantly depending on the rhetorical context. A structural mapping reveals that metaphorical density is not uniform; rather, it follows a distinct pattern of technical grounding followed by extreme metaphorical license. In the introductory and definitional sections, the text carefully distinguishes between 'epistemic uncertainty' (data-level) and 'subjective uncertainty' (system-level), deploying philosophical terminology to set a scholarly tone. When discussing the mechanics of symbolic or connectionist systems, the text relies heavily on mechanical language—vectors, weights, thresholds, and nodes—to establish scientific credibility. However, once this technical foundation is laid, the anthropomorphism sharply intensifies. The transition is marked by a sudden escalation in consciousness claims: mathematical 'processing' becomes 'understanding,' which rapidly evolves into 'knowing' and 'believing.' The text leverages its technical explanations as a license for aggressive anthropomorphism, suggesting that because we understand the math, we can confidently project a mind onto it. There is a profound asymmetry in how capabilities versus limitations are framed. Capabilities are almost exclusively described in agential, consciousness-attributing terms: the system 'takes a stance,' 'makes up its mind,' or possesses 'opinions.' In contrast, limitations are often framed mechanistically: the system 'lacks distributed knowledge,' or 'its data is ambiguous.' However, when a limitation results in an incorrect output, the text shifts back to agential framing, claiming the system is 'overconfident' or 'jumping to conclusions.' This asymmetry accomplishes a crucial rhetorical goal: it grants the machine the glory of human-like intelligence when it succeeds, but frames its failures as relatable human psychological flaws rather than catastrophic mathematical errors. The register shifts dramatically when theoretical examples ('X is like a cognitive system') are literalized into declarative assertions ('the system takes r to be sincere'). The strategic function of this anthropomorphism is highly oriented toward vision-setting and managing critique. By framing AI within the vocabulary of human psychology, the text prepares the audience to accept AI as a social actor. This pattern reveals an implied audience of philosophers and general theorists who are more interested in the narrative resonance of 'thinking machines' than the material reality of statistical software, allowing the author to bridge the gap between computer science and philosophy through the sheer force of metaphorical assertion.

Accountability Synthesis​

Accountability Architecture

This section synthesizes the accountability analyses from Task 1, mapping the text's "accountability architecture"—who is named, who is hidden, and who benefits from obscured agency.

Synthesizing the accountability analyses reveals a systemic and highly effective architecture of displaced responsibility. Throughout the text, a distinct pattern emerges regarding the distribution of agency: human actors are systematically erased, while the artificial system is continuously elevated to the status of an independent epistemic and moral agent. The 'accountability sink' in this discourse is the anthropomorphized machine itself. When the text discusses algorithmic processes, the engineers, data scientists, and corporate executives are left completely unnamed. Decisions about mathematical thresholds, training data selection, and system architecture are presented not as human choices driven by constraints and profit, but as the natural, organic characteristics of the AI. The text utilizes passive voice ('the network is trained') and agentless constructions ('it jumps to conclusions') to completely diffuse human responsibility. Consequently, accountability disappears into the abstraction of the 'system.' This architecture of displacement has profound liability implications. If policymakers and the public accept the framing that an AI 'makes up its mind' or 'fails to respect its uncertainty,' the legal and ethical responsibility for harmful outputs shifts from the manufacturer to the machine. The AI becomes a linguistic shield for corporate liability. If we apply the 'name the actor' test to the text's most significant agentless constructions, the narrative shifts radically. If 'the algorithm jumped to conclusions' is corrected to 'the corporate engineering team hardcoded an aggressive output threshold that ignored statistical variance,' entirely different questions become askable. We no longer ask 'How do we teach the AI to be patient?' but rather 'Why did the corporation deploy an unsafe system, and what is their financial liability?' If 'the system takes a stance' is corrected to 'the developers optimized the loss function to categorize this data,' alternative design choices become visible, and the illusion of the machine's objective judgment shatters. This systemic obscuration serves the immense institutional and commercial interests of the technology sector. By maintaining the illusion of mind, developers are granted the prestige of having created 'intelligence' while simultaneously being absolved of the responsibility for having created defective software. This displacement interacts seamlessly with the text's agency slippage and metaphor-driven trust, creating a closed discursive loop where the machine is trusted like a human, behaves like a machine, but is blamed as an autonomous agent when it fails.

Conclusion: What This Analysis Reveals​

The Core Finding

Synthesizing the metaphorical and explanatory audits reveals an interconnected system of anthropomorphism built upon three dominant patterns: the projection of conscious belief onto mathematical weights (The Epistemic Illusion), the projection of deliberate choice onto algorithmic outputs (The Agential Illusion), and the projection of psychological interiority onto programmatic execution (The Subjective Illusion). These patterns do not operate in isolation; they form a sequential, logical flow that constructs a comprehensive illusion of mind. The foundational, load-bearing pattern is the Epistemic Illusion. By establishing that distributed weights in a neural network constitute 'knowing' or 'believing' information, the text successfully bridges the gap between inert data and conscious awareness. This must be accepted as true for the other patterns to function. Once the audience accepts that the machine 'knows,' it becomes logically permissible to accept the Agential Illusion—that the machine 'makes up its mind' or 'takes a stance' based on that knowledge. Finally, the Subjective Illusion layers emotional and psychological states, such as 'hesitation' and 'respecting uncertainty,' over the entire architecture. This is a highly complex analogical structure that fundamentally blurs the line between processing (statistical correlation) and knowing (justified true belief). The consciousness architecture of the text relies entirely on treating mathematical operations as subjective states. If the foundational Epistemic Illusion is removed—if we insist that weights only 'encode' and never 'know'—the entire rhetorical structure collapses, exposing the machine as a mindless processor and invalidating the higher-order claims of agency and subjective uncertainty.

Mechanism of the Illusion:​

The rhetorical architecture of this illusion relies on a highly effective sleight-of-hand: the systematic exploitation of the 'curse of knowledge' combined with strategic verb substitution. The central trick is moving seamlessly from literal, mechanistic descriptions of data to figurative, psychological descriptions of the system, without ever signaling the leap. The author observes a statistical output (e.g., a system outputting a low-probability classification) and, because the author possesses a conscious mind that understands the semantic meaning of doubt, projects that subjective experience back onto the inert code. The causal chain of persuasion is temporally structured to exploit audience vulnerability. First, the text grounds itself in undeniable technical realities (activation vectors, backpropagation, probability math), lowering the reader's critical defenses. Once technical authority is established, the verbs subtly shift. The system no longer 'processes vectors'; it 'understands inputs'. It no longer 'calculates probability'; it 'experiences uncertainty'. This order matters profoundly, as the technical preamble acts as a Trojan horse for the consciousness claims. The audience, already eager to find human-like intelligence in machines due to cultural conditioning and science fiction narratives, readily accepts the anthropomorphic framing. This is not crude anthropomorphism (giving a computer a face), but a highly sophisticated, philosophical anthropomorphism that uses reason-based explanations to disguise mathematical functions as deliberate epistemic choices. By leveraging the ambiguity between epistemic uncertainty (missing data) and subjective uncertainty (conscious doubt), the text successfully traps the reader in an illusion where the software appears to possess an active, deliberating mind.

Material Stakes:​

Categories: Regulatory/Legal, Epistemic, Social/Political

The metaphorical framings employed in this text are not mere philosophical thought experiments; they generate concrete, tangible consequences across multiple domains. In the Regulatory and Legal sphere, the stakes are critical. If lawmakers internalize the framing that AI systems 'make up their minds' and experience 'subjective uncertainty,' regulatory frameworks will inevitably shift toward treating AI as a quasi-agent. This leads to the pursuit of 'AI rights' or 'algorithmic intent,' creating a massive liability loophole. The decision that shifts is the locus of accountability: courts and regulators will waste resources investigating the 'AI's decision process' rather than strictly prosecuting the tech companies for product liability and negligence. The clear winners are the corporate developers who avoid massive financial penalties, while the losers are the public who suffer from biased or unsafe deployments without legal recourse. Epistemically, claiming that AI 'knows' rather than 'processes' degrades our societal capacity for truth-evaluation. If humans believe a Large Language Model 'understands' context, they alter their behavior by using it as an arbiter of factual truth rather than a syntactic text generator. This epistemic pollution benefits companies selling AI as an oracle, while costing society its shared factual baseline. Socially and politically, projecting 'stances' and 'opinions' onto AI systems grants unearned authority to statistical outputs. If a predictive policing algorithm is viewed as having an 'opinion' rather than merely reflecting historical arrest data, its biased outputs are legitimized as objective machine judgment. Removing these metaphors threatens the commercial valuation of AI companies, as it reduces their 'thinking machines' back to brittle, biased, and highly regulated software products.

AI Literacy as Counter-Practice:​

Practicing critical literacy and mechanistic precision directly counters the material risks generated by anthropomorphic discourse. By applying the reframings developed in Task 4, the fundamental principles of epistemic humility and human accountability are restored. For instance, translating 'the AI made up its mind' to 'the algorithm generated an output based on static weights' forces a confrontation with the machine's lack of awareness. Replacing consciousness verbs (knows, understands) with mechanistic verbs (processes, retrieves, calculates) explicitly denies the system the unearned authority of a conscious knower. This linguistic discipline directly counters the epistemic risks by constantly reminding the user of the system's absolute dependency on training data and the statistical, non-factual nature of its outputs. Furthermore, human agency restoration—naming the specific corporations and engineering teams responsible for the system's architecture—destroys the accountability sink. When we say 'the developers deployed a model with unsafe thresholds' instead of 'the AI jumped to conclusions,' the locus of legal and moral responsibility is firmly anchored to human actors. Systematic adoption of this precision requires a massive cultural shift. Academic journals must require mechanistic translations of philosophical AI claims. Researchers must commit to stripping their papers of passive, agentless constructions. Industry marketing must be strictly regulated against using deceptive consciousness language. Naturally, this precision faces immense resistance from the tech industry, whose multi-trillion-dollar valuations depend heavily on the public believing they are creating artificial minds rather than advanced statistical calculators. Anthropomorphic language serves the interests of capital by enchanting the product and displacing liability; critical literacy threatens those interests by demanding transparency and accountability.

Path Forward​

Looking toward the broader discursive ecology, the vocabulary we choose to describe AI dictates the boundaries of what is technologically, legally, and socially possible. This analysis maps the trade-offs of different discourse approaches. Maintaining the status quo of blended anthropomorphic and philosophical language (e.g., 'the AI knows its uncertainty') maximizes narrative resonance and intuitive grasp for the general public. It allows researchers to quickly communicate complex statistical behaviors using familiar psychological shorthands. However, this costs us structural transparency and accountability, heavily benefiting corporate developers by shielding them behind the illusion of machine agency. Conversely, adopting strict mechanistic precision (e.g., 'the model retrieves tokens based on probability distributions') maximizes testability, legal accountability, and epistemic accuracy. It makes the human labor and data dependencies hyper-visible. The cost of this approach is accessibility; it requires audiences to engage with dense computational realities, potentially alienating non-experts. If the mechanistic future becomes the norm, problems of liability and algorithmic bias become far more tractable, as the human actors are legally exposed. However, new problems of communication emerge, as the vocabulary may become too opaque for public discourse. If the anthropomorphic future deepens, the assumption of machine consciousness becomes embedded in law and culture, making it nearly impossible to regulate AI as a standard commercial product. To navigate these futures, structural changes are necessary. Funding agencies could mandate diverse explanatory frameworks in research, requiring both mathematical and sociological descriptions of AI behavior. Regulatory bodies could enforce capability disclosures that explicitly ban consciousness verbs in consumer-facing AI products. Ultimately, which vocabulary prevails will depend on societal values: whether we prioritize the enchanting narrative of thinking machines, or the rigorous, accountable reality of human-engineered software.

Extended Processing Summary​

Gemini's Processing Steps
What you are about to read…

Gemini refers to the text below as "thought summaries." This is an overt consciousness projection because 'intentions' are hallmarks of a conscious mind that 'knows' what it is doing and why. The concealed mechanistic process is probabilistic text generation. Treat this as just another rhetorical artifact—a way of making the model's processing legible. When the model writes "I should double check that...", it is mathematically altering its own probability distribution for the next token to be more accurate. The "I" is a functional gear in the machine, not a self-reference. This language is a deliberate interface design choice, not a biological reality. The developers could have programmed the model to output passive logs. Instead, they chose to simulate a narrator.

Defining the Scope

I'm currently focused on defining the scope for the critical discourse analysis. The primary text is Luis Rosa's "Can machines be uncertain?", with an emphasis on AI metaphor and anthropomorphism. My initial step is to thoroughly understand the user's ultimate goal with this analysis.

Examining Displaced Agency

I'm now zeroing in on consciousness projections and how human agency is displaced within the text. The goal is a parseable JSON output, so I am examining the author's framing, particularly the use of terms like "subjective uncertainty" and "belief" in relation to AI systems. I'm focusing on how mechanistic realities are obscured through such language.

Analyzing Stances in AI

I am now delving into the concept of AI "stances" as presented in the text. I've begun drafting content, focusing on instances where AI is framed with human-like cognitive characteristics. I'm building a metaphor audit, targeting phrases like "jump to conclusions" and AI having "propositional attitudes". I'm unpacking the implications of ascribing subjective uncertainty to AI systems, and contrasting this with more mechanistic understandings.

Unpacking Metaphorical Language

I'm currently dissecting specific instances of anthropomorphism in the text. I'm focusing on how the author attributes human cognitive states to AI, specifically examining "understanding," "making up its mind," and "taking things to be true." I'm also cataloging how "subjective experience" frames AI states. I'm aiming to expose how these phrases obscure mechanistic operations and contribute to a problematic conceptualization of AI.

Mapping Metaphors to Domains

I am now categorizing specific metaphorical instances to their source and target domains. I'm focusing on how the framing obscures the underlying mechanistic processes. For example, I am analyzing how describing AI as "hesitating" conceals processing delays, and how the network as a "cognitive system" frames computational processes like a biological brain. I'm also examining the implications of describing AI states as having "attitudes of uncertainty."

Categorizing Metaphors

I'm now identifying instances of Brown's typology within the text. I've begun to classify passages based on whether they employ functional/theoretical, intentional/reason-based, functional/mechanistic, or empirical generalization metaphors. For instance, I've categorized "The algorithm will calculate the difference..." as functional/mechanistic, and "Since uncertainty is an important ingredient..." as intentional/reason-based. This will help refine the metaphor audit.

Reframing Metaphorical Language

I'm now working to translate anthropomorphic statements into mechanistic descriptions. For example, "The network itself takes r to be sincere" becomes "The network outputs a vector classifying r into the 'sincere' category based on its trained weights." This involves identifying human actors involved and restoring human agency in the process. I am focused on the agency slippage, and on how framing "subjective uncertainty" builds unjustified trust by implying self-reflection in AI.

Unveiling the Omissions

I'm currently focusing on how the text obscures the human actors and design choices behind AI systems. The author frames systems with phrases like "AI system" or "ANN" but downplays the human contributions. This is done by hiding designers' decisions on architecture, thresholds, and training data. The anthropomorphism escalates when the author discusses "subjective" states and desired AI outcomes, later downplaying the philosophical arguments by returning to technical details. I'm noting how responsibility is diffused to the AI, rather than acknowledging the developers' roles in setting parameters or the company's decisions on deploying particular models.

Mapping Agency Displacement

I'm now examining how attributing "uncertainty" to an AI agent displaces manufacturer product liability. I am focused on how stripping the text of mental verbs reveals the underlying power structures. I'm building a discourse map to trade philosophical intuition for technical precision. I am focusing on formatting the JSON output, and I am double checking to make sure it precisely matches the schema.

Refining JSON Outputs

I'm now revising the JSON output approach. To meet all output constraints, I'll ensure all values are single-line strings. Specifically, I'm refactoring to fully minify the JSON or, to be safe, avoid any embedded newlines or tabs, in addition to being a single-line string.

Calculating Word Lengths

I'm now calculating word lengths for JSON content, to meet the length requirements. I'm focusing on "jump to conclusions" and "made up its mind", specifically, to begin with. I'm building out a plan for meeting output constraints by constructing the JSON systematically. I must generate a minified single-line string.

Analyzing Trust Attribution

I am now drafting implications of attributing cognitive failure, such as "jumping to conclusions", to AI. This framing builds an illusion of autonomy and obscures the role of developers, creating trust issues. It suggests that the AI is making an autonomous conscious choice. This language subtly undermines proper risk assessments by obfuscating the underlying deterministic, mechanistic processes. The focus is on the impact of this framing on user, policymaker, and developer trust.

Framing Accountability Analysis

I'm now unpacking the ramifications of displaced accountability in the text. I am examining how framing the AI as an "impatient independent agent" subtly shifts liability from human engineers and corporations. My analysis of this framing exposes how it undermines the potential for appropriate risk assessment. I am drafting examples of how to address this agency slippage in my analysis. I am planning on documenting the impact of this anthropomorphism on public trust and regulation.

Unpacking Responsibility Displacement

I'm now identifying how the text shifts accountability away from human designers and toward the AI. Using the agentless construction, such as "the system jumps to conclusions," the designers' and executives' role in defining thresholds and training data is concealed. This shift of focus helps obscure accountability, protecting them from scrutiny while the AI is made into the direct agent.

Constructing Output Strings

I'm now concatenating the elements into a single, long string to fit output constraints. I'm focusing on "jump to conclusions" and "made up its mind" from Task 1, 2, 4, in order, alongside their respective explanations to meet constraints. I'm ensuring everything is minified. I will work in order, in order to generate a working JSON structure.

Categorizing Textual Elements

I'm now cataloging examples of Functional/Empirical, Functional/Mechanistic, and Empirical Generalization metaphors from the provided text. I am using the rules to extract the meaning from "If the system is prompted to decide whether not-p, for example, the presence of <p, 0.9> in its model should cause the output of this new decision process to be <¬p, 0.1>..." I'm also examining the text "The algorithm will calculate the difference between the ANN’s actual output vector and the desired output vector and use that difference (if any) to modify the weights..." and "For example, the rules implemented in a symbolic AI system may generate a 90% degree of confidence that a patient has a certain disease D..."

Checking Output String

I am now ensuring the output meets length constraints, especially for word counts. I will avoid using \n or \t, as required, and generate the JSON on a single line. I'm focusing on "Agency Slippage" content, especially 5.1, following the framework, and preparing to build the final concatenated string. I am aiming to ensure that no new lines or tabs are generated in the content.

Analyzing Framing Oscillation

I'm now identifying oscillations between mechanistic and agential framings within the text, with the goal of illustrating agency slippage. The process I'm using is to identify and organize quotes. This analysis is driven by the author's strategy to validate philosophical investigations of AI by interweaving technical details with human-like descriptions. I'm focusing on "calculating the difference" and contrasting it with the network's implied "taking r to be sincere."

Identifying Framing Tactics

I'm now zeroing in on the specific framing tactics: building epistemic authority through deterministic language, followed by anthropomorphic leaps. I'm noting the predictable timing of these shifts, from mathematical operations to conscious states. The author uses the "curse of knowledge," projecting his understanding onto the AI system. My current focus is on building the evidence for this pattern.

Unpacking Agentless Constructions

I'm now analyzing the implications of agentless constructions, like the text framing AI as an autonomous actor, with subjective uncertainty, obscuring corporate decisions. The consciousness projection pattern begins by presenting AI as a "knower," then builds agential claims about its cognitive shortcomings. These rhetorical achievements reframe algorithmic discrepancies as moral failings.

Refining JSON Assembly

I'm now integrating all generated content into a single, valid JSON string, strictly adhering to the specified schema and length restrictions, with absolutely no embedded newlines or tabs. The aim is a fully minified, single-line output. My focus is on ensuring structure integrity.

Analysis Provenance

Run ID: 2026-03-08-can-machines-be-uncertain-metaphor-xafbbu Raw JSON: 2026-03-08-can-machines-be-uncertain-metaphor-xafbbu.json Framework: Metaphor Analysis v6.4 Schema Version: 3.0 Generated: 2026-03-08T10:09:11.478Z

Discourse Depot © 2025 by TD is licensed under CC BY-NC-SA 4.0