The Rhetoric of Machine Learning

arxiv: 2604.06754 · v1 · submitted 2026-04-08 · 💻 cs.LG · cs.CY

The Rhetoric of Machine Learning

Robert C. Williamson This is my paper

Pith reviewed 2026-05-10 17:48 UTC · model grok-4.3

classification 💻 cs.LG cs.CY

keywords machine learningrhetoricpersuasionmanipulation as a serviceworld modelsbusiness modelsobjectivity

0 comments p. Extension

The pith

Machine learning is inherently rhetorical rather than a neutral and objective way to build world models from data.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper argues that machine learning should be understood through the lens of rhetoric, the art of persuasion, instead of as a neutral technology for extracting objective models from data. This matters because it shifts attention to how ML systems influence beliefs and actions in practice, especially in commercial settings. The author explores specific rhetorical aspects of the technology and focuses on the business model of manipulation as a service as a key illustration of this persuasive character.

Core claim

Rather than being a neutral and objective way to build world models from data, machine learning is inherently rhetorical. The paper examines some of its rhetorical features and examines one pervasive business model where machine learning is widely used, manipulation as a service.

What carries the argument

The rhetorical perspective on machine learning, which reframes the technology as persuasion rather than neutral modeling and applies this to both technical features and commercial applications.

If this is right

Machine learning systems would be evaluated in part on the persuasive effects they produce rather than solely on predictive accuracy.
The manipulation as a service model would be treated as a central and expected use of the technology rather than an aberration.
Design decisions in machine learning would be recognized as choices about what to persuade users to believe or do.
Ethical and regulatory scrutiny of machine learning would need to address its persuasive dimensions directly.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

This perspective implies that standards for transparency in machine learning should include disclosure of intended persuasive goals.
It connects machine learning to longer traditions of analyzing media and technology as forms of argument.
Developers might adopt practices from rhetoric and communication ethics when building systems.

Load-bearing premise

That the rhetorical lens captures the essential nature of machine learning technology rather than being one interpretive frame among many.

What would settle it

A concrete demonstration of a widely deployed machine learning system whose design, training, outputs, and business use contain no element of persuasion or influence on beliefs or actions.

read the original abstract

I examine the technology of machine learning from the perspective of rhetoric, which is simply the art of persuasion. Rather than being a neutral and "objective" way to build "world models" from data, machine learning is (I argue) inherently rhetorical. I explore some of its rhetorical features, and examine one pervasive business model where machine learning is widely used, "manipulation as a service."

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper makes a reasonable case for seeing machine learning through the lens of rhetoric but doesn't establish it as the primary or only useful frame.

read the letter

The one or two things to know are that the paper presents machine learning as an inherently rhetorical practice focused on persuasion, and it uses the business model of selling manipulation to illustrate the point. Williamson applies concepts from classical rhetoric to examine how ML systems are constructed and deployed. He argues that rather than building objective world models, these systems are designed to influence outcomes and behaviors. The paper does this well by identifying rhetorical features in data selection, model objectives, and output presentation. It also examines manipulation as a service as a widespread approach in industry, which brings the abstract idea down to concrete commercial practices. This approach gives a fresh way to think about why ML often prioritizes user engagement over other goals. The softer areas are around the strength of the inherently claim. While the rhetorical lens highlights persuasive elements effectively, machine learning also serves non-persuasive purposes in research and engineering where the goal is accurate forecasting without intent to change minds. The paper does not provide detailed counterexamples or quantitative analysis to show how dominant the rhetorical aspect is across the field. As a conceptual piece, it relies on logical argument and examples rather than new evidence, which is appropriate but means the conclusions depend on how convincing the reader finds the framing. This work is aimed at scholars and practitioners in AI ethics, philosophy of technology, and policy who are already questioning the neutrality of algorithms. A reader looking for algorithmic improvements or empirical studies will find less to take away. I would recommend that a serious editor send this to peer review. The ideas are well-articulated and timely, and referees could help refine the connections to existing literature or suggest additional illustrations without undermining the core perspective.

Referee Report

0 major / 2 minor

Summary. The manuscript argues that machine learning is inherently rhetorical (the art of persuasion) rather than a neutral, objective method for building world models from data. It examines rhetorical features of ML and analyzes one common application domain through the lens of a 'manipulation as a service' business model.

Significance. If the interpretive framing is coherent and insightful, the paper could contribute to broader discussions on the non-neutral aspects of ML systems, particularly their role in influence and persuasion, thereby informing ethical and societal analyses of the technology.

minor comments (2)

The abstract states the central claim clearly but does not preview the specific rhetorical features to be explored; adding one or two concrete examples here would improve reader orientation without altering the perspective-piece format.
Terminology such as 'rhetorical features' and 'inherently rhetorical' is used repeatedly; a brief definitional paragraph early in the manuscript would reduce potential ambiguity for readers outside rhetoric studies.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their summary of the manuscript and for noting its potential contribution to discussions on the non-neutral aspects of ML systems. We appreciate the recommendation for minor revision.

Circularity Check

0 steps flagged

No significant circularity in interpretive argument

full rationale

The paper is a conceptual perspective piece arguing that machine learning is inherently rhetorical. It advances no mathematical derivations, equations, fitted parameters, predictions, or formal proofs. The central claim is presented explicitly as an interpretive framing to be explored through examples and features rather than as a technical result derived from prior inputs. No self-citations, ansatzes, or uniqueness theorems are invoked in a load-bearing way that reduces the argument to its own definitions or fitted quantities by construction. The derivation is therefore self-contained as a non-technical essay.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The paper relies on an interpretive framework rather than formal axioms, free parameters, or new entities. The central claim rests on treating rhetoric as the appropriate analytical lens for ML.

axioms (1)

domain assumption Machine learning technology is best understood as a form of persuasion rather than neutral modeling.
This interpretive premise underpins the entire examination of rhetorical features and business models.

pith-pipeline@v0.9.0 · 5337 in / 1098 out tokens · 54574 ms · 2026-05-10T17:48:37.727750+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

2 extracted references · 2 canonical work pages · 1 internal anchor

[1]

Language Modeling Is Compression

Many LLMs do not declare the data they were trained upon (or like Llama, just refer to the “pile”, an unspecified collection of pirated texts.) Stephen Toulmin, The Uses of Argument, Cambridge University Press, 2003.31 That science requires the whole chain (or better said, network or tangle) is nicely argued in Nancy 32 Cartwright, Jeremy Hardie, Eleonora...

work page internal anchor Pith review arXiv 2003
[2]

Writing is a technology that restructures thought,

See also his “Writing is a technology that restructures thought,” pages 293—319 in The Linguistics of Literacy, John Benjamins Publishing Company, 1992. The full story of Ramus is the subject of the delightful book by Walter J. Ong, Ramus, method, and the 62 decay of dialogue: From the art of discourse to the art of reason. University of Chicago Press, 20...

work page 1992

[1] [1]

Language Modeling Is Compression

Many LLMs do not declare the data they were trained upon (or like Llama, just refer to the “pile”, an unspecified collection of pirated texts.) Stephen Toulmin, The Uses of Argument, Cambridge University Press, 2003.31 That science requires the whole chain (or better said, network or tangle) is nicely argued in Nancy 32 Cartwright, Jeremy Hardie, Eleonora...

work page internal anchor Pith review arXiv 2003

[2] [2]

Writing is a technology that restructures thought,

See also his “Writing is a technology that restructures thought,” pages 293—319 in The Linguistics of Literacy, John Benjamins Publishing Company, 1992. The full story of Ramus is the subject of the delightful book by Walter J. Ong, Ramus, method, and the 62 decay of dialogue: From the art of discourse to the art of reason. University of Chicago Press, 20...

work page 1992