The Rhetoric of Machine Learning
arXiv:2604.06754v1 Announce Type: new 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."
arXiv:2604.06754v1 Announce Type: new 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."
Executive Summary
The article 'The Rhetoric of Machine Learning' critically reframes machine learning (ML) not as an objective scientific endeavor but as an inherently rhetorical art of persuasion. It posits that ML systems are designed to influence, rather than merely represent, reality, challenging the prevalent notion of their neutrality. The author explores specific rhetorical features embedded within ML technologies and introduces 'manipulation as a service' as a pervasive business model that leverages ML's persuasive capabilities. This perspective offers a valuable departure from purely technical analyses, urging a deeper socio-ethical examination of ML's design, deployment, and impact on individual and societal decision-making.
Key Points
- ▸ Machine learning is fundamentally rhetorical, actively persuading rather than passively modeling reality.
- ▸ The notion of ML as a neutral or objective technology is challenged, highlighting its inherent persuasive design.
- ▸ The article explores specific rhetorical features embedded within ML systems.
- ▸ It identifies 'manipulation as a service' as a significant business model driven by ML's persuasive power.
Merits
Novel Conceptual Framework
The application of rhetorical theory to machine learning offers a fresh and insightful lens, moving beyond standard ethical or technical critiques to a more fundamental understanding of ML's persuasive nature.
Challenges Dominant Narratives
Effectively deconstructs the 'objectivity' myth surrounding ML, which is crucial for fostering more critical engagement with AI technologies.
Highlights Business Model Implications
The concept of 'manipulation as a service' provides a concise and potent descriptor for a significant, often unacknowledged, aspect of the ML economy.
Demerits
Scope of Rhetorical Features
While mentioning 'some rhetorical features,' the abstract does not indicate the depth or breadth of their exploration, potentially leaving the reader wanting more concrete examples or a typology.
Specificity of 'Manipulation'
The term 'manipulation' can be broad; a more nuanced breakdown of different types or degrees of manipulation facilitated by ML would strengthen the argument.
Lack of Counterarguments/Nuance
The abstract presents a strong thesis but doesn't hint at acknowledging situations where ML might genuinely aim for objectivity, even if imperfectly, or where its persuasive aspects are benign.
Expert Commentary
This article's central thesis – that machine learning is fundamentally rhetorical – represents a significant and timely contribution to the interdisciplinary discourse surrounding AI. By shifting the analytical paradigm from technical objectivity to persuasive intent, the author compels a deeper interrogation of ML's societal role. The concept of 'manipulation as a service' is particularly potent, offering a concise framework for understanding prevalent business models that leverage ML to shape user behavior and preferences. From a legal and policy perspective, this reframing is critical. If ML systems are inherently persuasive, then their impact cannot be merely assessed through metrics of accuracy or efficiency, but must be scrutinized for their influence on autonomy, consent, and democratic processes. This necessitates a re-evaluation of existing regulatory instruments, which largely assume a neutral technological substrate. Future scholarship should elaborate on specific rhetorical techniques employed by ML, providing a taxonomy that informs both technical design and regulatory foresight. The article lays crucial groundwork for moving beyond superficial ethical checklists to a profound engagement with ML's persuasive power.
Recommendations
- ✓ Expand on the specific rhetorical features of machine learning, providing detailed examples and a categorization framework.
- ✓ Explore the ethical implications of 'manipulation as a service' across various sectors (e.g., finance, health, politics) and propose potential ethical safeguards.
- ✓ Analyze the historical and philosophical roots of persuasion in technology to contextualize ML's rhetorical nature.
- ✓ Propose a framework for auditing the 'rhetorical intent' and 'persuasive impact' of machine learning algorithms and systems.
Sources
Original: arXiv - cs.LG
