When Specs Lie Or, The maths your PRD is failing

Example output (abridged):

FORMAL LOGICAL ANALYSIS Article: "It's Time to Talk About Ethics" by Mcauldronism Date of Analysis: January 2025

================================================================================ 1. MAIN CONCLUSION ================================================================================

The article argues for a compound conclusion:

PRIMARY CLAIM: Whether we recognize tool-assisted cognition (including AI-assisted cognition) as "genuine" cognition is fundamentally an ethical choice, not merely a factual determination.

SECONDARY CLAIM: The morally correct choice is to recognize tool-assisted cognition as genuine—and those who deny it are committing a moral failure analogous to denying that wheelchair users are "really" mobile.

================================================================================ 2. ARGUMENT STRUCTURE ================================================================================

The argument proceeds through analogical reasoning, moving from uncontroversial cases to the contested case. Here is the logical flow:

STAGE 1: Setup (The Cringe) - Author's initial resistance to ethical framing of extended cognition - Transition to accepting Andy Clark's view that ethics is unavoidable here

STAGE 2: The Wheelchair Analogy - Premise: A wheelchair user is genuinely mobile - Premise: The wheelchair is part of how they move - Premise: Denying this makes one "an asshole" (moral judgment) - Implied principle: Functional integration = genuine attribution

STAGE 3: The Otto Case - Premise: Otto (Alzheimer's patient) uses a notebook to remember - Premise: The notebook functions as part of his memory system - Premise: Denying this is "pedantic at best, ableist at worst" - Application of same principle from Stage 2

STAGE 4: Extraction of Moral Principle - These judgments reveal an underlying choice about how we see tool use - This choice is ethical, not merely descriptive

STAGE 5: Application to AI - If the principle holds for wheelchairs and notebooks, it holds for AI - Someone achieving something with AI is genuinely achieving it - Denying this would be morally analogous to denying wheelchair mobility

STAGE 6: Personal Disclosure - Author acknowledges writing with Claude - Frames this as "extended cognition" consistent with the argument

STAGE 7: Call to Action - Reader must make a choice - That choice reveals their values about human nature

================================================================================ 3. FORMAL RECONSTRUCTION ================================================================================

KEY FOR SYMBOLS: - Px = "x is a person" - Tx = "x is a tool" - F(x,t) = "x uses tool t in a functionally integrated way" - A(x,t,φ) = "x genuinely performs activity φ using tool t" - D(x,t,φ) = "x denies that someone genuinely performs φ when using t" - M(x) = "x commits a moral failure"

CORE PRINCIPLE (Parity Principle, implicit): ∀x∀t∀φ[(Px ∧ Tx ∧ F(x,t)) → A(x,t,φ)]

In natural language: For any person and any tool, if the person uses the tool in a functionally integrated way, then they genuinely perform the relevant activity with that tool.

MORAL PRINCIPLE (implicit): ∀y∀x∀t∀φ[(Py ∧ D(y,x,t,φ) ∧ F(x,t)) → M(y)]

In natural language: Anyone who denies genuine attribution when functional integration exists commits a moral failure.

THE ARGUMENT FORMALIZED:

P1: F(wheelchair-user, wheelchair) — wheelchair is functionally integrated P2: A(wheelchair-user, wheelchair, mobility) — therefore genuine mobility P3: F(Otto, notebook) — notebook is functionally integrated P4: A(Otto, notebook, remembering) — therefore genuine remembering P5: F(AI-user, AI) — AI can be functionally integrated --- C1: A(AI-user, AI, thinking/creating) — therefore genuine cognition/creation

MORAL EXTENSION:

P6: D(critic, wheelchair-user, wheelchair, mobility) → M(critic) P7: D(critic, Otto, notebook, remembering) → M(critic) P8: The same principle applies uniformly to all functionally integrated tools --- C2: D(critic, AI-user, AI, creating) → M(critic)

================================================================================ 4. VALIDITY ASSESSMENT ================================================================================

LOGICAL FORM: The argument is structurally valid IF we grant: (a) The analogies hold (wheelchair ≈ notebook ≈ AI in relevant respects) (b) The parity principle is true (c) Moral judgments transfer across analogous cases

TYPE OF REASONING: - Primarily analogical (cases → principle → new case) - Secondarily deductive (once the principle is established)

VALIDITY VERDICT: The argument is valid in its FORM. The conclusion follows from the premises if we accept them. However, the soundness depends entirely on whether the analogies are apt and whether the parity principle holds for AI.

The move from "functional integration" to "genuine attribution" is the crux. This is philosophically contested (see Section 6).