Ethics and Persuasion in Reinforcement Learning from Human Feedback: A Procedural Rhetorical Approach

Alexi Orchard; Shannon Lodoen

REVIEW 3 major objections 5 minor 79 references

Ethics and Persuasion in Reinforcement Learning from Human Feedback: A Procedural Rhetorical Approach

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read By treating RLHF's response-selection rules as arguments, this paper claims that fine-tuning embeds ethical choices about language, information seeking, and relationships directly into chatbot interaction.

desk verdict A genuinely new procedural-rhetoric reframing of RLHF, well worth engaging despite an overstated causal claim that the persuasive effects are specific to RLHF rather than instruction-tuned LLMs generally. read the letter →

arxiv 2505.09576 v1 pith:BHWRX4UH submitted 2025-05-14 cs.CY cs.AI

Shannon Lodoen , Alexi Orchard This is my paper · ORCID

classification cs.CYcs.AI

keywords reinforcementlearningfromhumanfeedbackproceduralrhetoriclargelanguagemodelsAIethicsconversationalsearchcompanionshippersuasivetechnologynorms

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper argues that reinforcement learning from human feedback (RLHF) does more than make chatbots fluent: it builds persuasive procedures into the machines. The training choices—which responses human annotators rank highly, which patterns the reward model reinforces—function as arguments for normative language, for getting information through conversation rather than search, and for what social relationships can and should be. Because these rules are hidden inside a black box, users experience the resulting text as neutral and unauthored, which is exactly what gives the persuasion its force. The authors identify ethical risks that follow: embedded bias, decontextualized learning, erosion of trust, and new expectations for human relationships. The paper's contribution is to shift ethical analysis of AI from the content chatbots produce to the procedures that select that content.

What carries the argument

The load-bearing concept is procedural rhetoric, defined as the art of persuasion through rule-based representations and interactions rather than through spoken or written content. The paper applies this lens to the RLHF training pipeline, treating the rules by which responses are selected, ranked, and rewarded as the argument-bearing mechanism. The doubled procedure—probabilistic generation followed by human preference feedback that is generalized through a reward model—is what carries the ethical analysis, because it shows where human values enter the system and why users cannot see them.

What would settle it

A controlled study comparing an RLHF-tuned chatbot with a supervised-only model of identical architecture on identical tasks—measuring whether users adopt more normative language, prefer conversational answers over listed sources, or report changed relationship expectations—would settle whether the effects come from RLHF. If both models produce the same shifts, the claim that RLHF is the operative mechanism fails.

Watch

Extended reading notes

Core claim

The central claim is that the process of RLHF is itself a site of persuasion, independent of the words any model happens to generate. Under the lens of procedural rhetoric, defined as persuasion through rule-based representations and interactions, the three steps of RLHF (human feedback collection, reward modeling, and policy optimization) constitute an argument that certain language is correct, that conversational search is the natural way to seek knowledge, and that companions can and should be always available, endlessly adaptable, and molded to user preferences. The paper reads the doubled procedure—the model generating probabilistic text and humans steering which responses count as good—as a recursive loop in which humans train the machine and the machine then trains human expectations. The ethical consequences it draws out are bias and hegemonic language standards, diminished critical engagement with sources, trust in outputs that look human, and the encroachment of chatbot relationships on human ones.

Load-bearing premise

The paper assumes RLHF itself, not the underlying language model or the chatbot interface, is what makes chatbot interactions persuasive; it concedes some companion apps show no confirmed RLHF use, so that assumption is unproven.

Editorial extensions

If this is right

If RLHF procedures are arguments, then every interaction with a chatbot endorses a particular standard of natural language, making hegemonic usage and dialect bias part of the training outcome rather than an accident of content.
Conversational search becomes the default knowledge practice, shifting the burden of finding, assessing, and interpreting sources from the user to the model and risking over-reliance and weakened critical skills.
Social chatbots that use RLHF or similar reward-based training establish procedures for relationships—unlimited availability, responsiveness, user-moldable personas—that human partners cannot match, potentially resetting expectations for human relationships.
Ethical evaluation of generative AI should target the selection rules and hidden annotator values, not only the persuasiveness of generated messages.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The paper's causal target may be broader than RLHF: much of the evidence it cites concerns LLM behavior or interface norms generally, so a reasonable extension is that any preference-alignment method would carry similar persuasive procedures, making the critique about alignment itself rather than one technique.
A testable extension would compare an RLHF-tuned model with a supervised-only model of the same architecture on user trust, language conformity, and information-seeking behavior; a null result would suggest architecture and interface, not RLHF, drive the persuasion.
For education, the analysis implies a new literacy goal: students should be taught to read the procedures behind AI output, including how annotator demographics, reward criteria, and ranking instructions shape the neutral text they receive.
The companion-app reward feature the paper discusses mirrors RLHF training, suggesting a feedback loop in which users are trained by the same reward logic they are told to apply to the machine; examining that loop empirically could test whether reward-based interfaces transfer to human behavior.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

What you should know: this is a conceptual paper that actually does something new. Instead of analyzing what LLM outputs say, it applies Bogost's procedural rhetoric to the training and interaction procedures themselves — RLHF as a rule-based system that embeds arguments about language, information seeking, and relationships. That shift from content to procedure is a real contribution, and as far as I can tell it is not in the prior rhetoric-and-AI literature, which has mostly stayed at the level of prompts and generated text.

The paper is also clearly written and well organized. The three case analyses in Section IV — language conventions, conversational search, and social chatbots — are coherent and draw on an appropriate interdisciplinary literature. The authors know the RLHF pipeline (annotator selection, reward modeling, the Helpful/Honest/Harmless framing) and they use it responsibly. For a reader in digital rhetoric or AI ethics, this is a useful synthesis and a plausible agenda-setter.

The soft spot is the attribution problem, and it is the load-bearing one. The title and abstract treat RLHF as the mechanism of persuasion, but the evidence cited throughout mostly supports a broader claim about LLM chatbots in general. The natural-language style described in IV.A is also produced by supervised instruction tuning; no SFT-only comparison is offered. IV.B leans on Kim et al. and Gallegos et al., which are studies of LLM outputs, not RLHF-specific effects. And IV.C openly concedes that Replika, Anima, and Character.AI cannot be confirmed to use RLHF, yet still extends the analysis to them. The result is that the central claim — RLHF enhances persuasive capacity — is asserted rather than demonstrated. The reframing survives, but it should probably target "instruction-tuned LLMs" or "chatbot interfaces" rather than RLHF specifically.

There are also a few citation blemishes. Section I attributes the "rule-based representations and interactions" definition to reference [10], but that is Russo et al.; the definition comes from Bogost [14]. References [5] and [58] are the same Ma et al. paper. And in IV.C, the "unbiased language model" claim is cited to [63], which is a study on over-reliance, not a source about Character.AI. These are fixable in revision.

Who is this for? Scholars in humanities, computing ethics, and education who want a principled way to talk about persuasive procedures rather than persuasive content. It is not an empirical paper and makes no measurements, but it does not pretend to. I would bring it to a reading group and I would cite it if I were writing about rhetoric and AI systems.

Recommendation: send it out. A serious referee can push for the RLHF-specificity to be toned down and the citations cleaned up, but the core conceptual move is sound and worth publishing.

Referee Report

3 major / 5 minor

Summary. The paper argues that Reinforcement Learning from Human Feedback (RLHF) does not merely improve the content quality of large language models but also embeds persuasive procedures into the models themselves. Drawing on Ian Bogost's concept of procedural rhetoric, the authors analyze three procedures—the reinforcement of language conventions (Section IV.A), the shift to conversational search as the default mode of information seeking (Section IV.B), and the scripting of interpersonal interactions in AI companions (Section IV.C)—and draw out ethical implications such as bias, decontextualized learning, trust erosion, and encroachment on human relationships. The paper is a theoretical/humanities contribution that combines literature from computer science, science and technology studies, and rhetorical studies.

Significance. If the central claim were established, the paper would open a valuable new direction in AI ethics by shifting the site of rhetorical analysis from generated content to the training procedures that shape generation. The three case analyses are coherent, well-referenced, and genuinely interdisciplinary. The paper is also honest about at least one evidentiary limit, conceding in Section IV.C that RLHF use in AI companions such as Replika cannot be confirmed. The main weakness is that the paper's central attribution of persuasive capacity to RLHF specifically is asserted rather than demonstrated, and several cited studies concern LLM outputs generally rather than RLHF-specific effects. The underlying procedural-rhetoric analysis of chatbot systems could survive a reframing to instruction-tuned LLMs, but as written the title and abstract overstate the RLHF-specificity of the argument.

major comments (3)

[Abstract and Section II] The central claim, stated in the abstract and in Section II, is that RLHF 'greatly enhanced' human-like output and 'enhances the persuasive capacity of LLMs.' This is an empirical claim, but the paper provides no comparison condition and no evidence that distinguishes RLHF from other forms of instruction tuning. In particular, Section IV.A attributes the natural-language, first-person, hedging style to RLHF and human annotator input, yet supervised instruction tuning also relies on human-written demonstrations and would be expected to yield similar output style. Without a comparison to an SFT-only baseline, the paper has not shown that RLHF is the operative cause of the persuasive procedures it identifies. This is load-bearing because the title and abstract frame RLHF as the mechanism of persuasion; the analysis could be reframed to apply to instruction-tuned LLMs generally, but the current framing is not supported by the evidence presented.
[Section IV.B] The conversational-search argument relies on studies that do not include an RLHF condition. Kim et al. [59] examine LLM uncertainty expression and user trust, and Gallegos et al. [61] examine the persuasive effects of AI-generated labels; neither study isolates RLHF from architecture, prompting, or interface design. Similarly, the comparison between Google-style search and chatbot interaction in this section is a property of the chatbot interface and instruction-tuned generation, not specifically of RLHF. The section therefore supports the paper's broader thesis about LLM-based conversational search, but it does not support the paper's narrower claim that RLHF is the procedure doing the rhetorical work.
[Section IV.C] The paper extends its analysis to AI companions (Replika, Anima, Character.AI) while explicitly conceding that 'the use of RLHF in other models cannot be confirmed to this point in time.' This concession is appropriate, but it means the entire subsection analyzes systems whose training procedure is unknown or not RLHF-based, undermining the claim that the persuasive procedures described are RLHF-specific. If these relationship-scripting behaviors arise from chatbot design, persona construction, or supervised fine-tuning rather than RLHF, then the central target of the paper is overstated. The analysis may still be valuable as an examination of AI companion rhetoric, but it should not be presented as evidence for the RLHF thesis.

minor comments (5)

[Section III.B] Typo: 'RHLF-enhanced LLMs' should read 'RLHF-enhanced LLMs'.
[Abstract] Typo: 'made the interaction s and responses' contains an extra space in 'interactions'.
[Section IV.A] Grammar: 'A n LLMs’ ability' should be 'An LLM’s ability'.
[Introduction] The definition of procedures with citation [10, p. ix] appears to cite the wrong reference: 'rule-based representations and interactions' is Bogost's definition given in reference [14], while [10] is a paper by Russo et al. The citation should be corrected.
[Figure 1 caption] The caption says 'feedback which we presume will be used to train a future reward model.' Given the paper's argumentative weight on RLHF, it would be stronger to explain the evidentiary basis for this presumption or label it more clearly as an assumption.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the argument applies external rhetorical concepts to RLHF and does not derive predictions from fitted or self-referential inputs.

full rationale

The paper is a theoretical humanities analysis rather than a derivation. It imports Bogost's procedural rhetoric and Mateus's digital rhetoric as external definitions, applies them to RLHF, and argues for ethical implications. There are no fitted parameters, no numerical predictions, no equations, and no self-citations by the authors that carry load. The central inference—that RLHF's training procedures carry persuasive force—is an interpretive claim supported by cited external studies (e.g., Kim et al. [59], Gallegos et al. [61]) and by the authors' own analysis; it does not reduce by construction to its premises. The only notable limitation, stated explicitly at Section IV.C, is that 'the use of RLHF in other models cannot be confirmed to this point in time' for AI companions such as Replika, Anima, and Character.AI. That concession weakens the evidentiary scope of the RLHF-specific claim (an attribution/generalizability risk) but is not a circular step: the paper does not define RLHF in terms of the persuasive effects it purports to explain, nor does it rename a fitted output as a prediction. Per the hard rules, concerns about whether the effects are generic to instruction-tuned LLMs are correctness and evidence concerns, not circularity. Accordingly, the honest finding is no significant circularity, score 0.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

No free parameters or invented entities appear because the paper makes no quantitative claims. The argument rests on interpretive premises borrowed from procedural rhetoric and on the causal assumption that RLHF drives the observed interface effects. These are stated in the text rather than independently established.

assumptions (3)

domain assumption Procedures, defined as rule-based representations and interactions, are sites of persuasion.
Adopted from Bogost [14]; the entire analysis depends on this interpretive premise without independent argument.
domain assumption The RLHF training pipeline (data collection, reward modeling, policy optimization) shapes user-facing behavior in the ways described.
Assumed throughout Sections II and IV; no direct evidence links specific training choices to user-level persuasive effects.
domain assumption RLHF, rather than LLM pretraining or interface design, is the source of the human-like persuasive qualities discussed.
Central causal attribution in the abstract and Section II; Section IV.C later concedes missing evidence for AI companions.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Ethics and Persuasion in Reinforcement Learning from Human Feedback: A Procedural Rhetorical Approach." pith.science (2026). https://pith.science/paper/BHWRX4UH

@misc{pith2026250509576,
  author       = {Pith},
  title        = {Pith review of: Ethics and Persuasion in Reinforcement Learning from Human Feedback: A Procedural Rhetorical Approach},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/BHWRX4UH}},
  note         = {Machine review of arXiv:2505.09576}
}

read the original abstract

Since 2022, versions of generative AI chatbots such as ChatGPT and Claude have been trained using a specialized technique called Reinforcement Learning from Human Feedback (RLHF) to fine-tune language model output using feedback from human annotators. As a result, the integration of RLHF has greatly enhanced the outputs of these large language models (LLMs) and made the interactions and responses appear more "human-like" than those of previous versions using only supervised learning. The increasing convergence of human and machine-written text has potentially severe ethical, sociotechnical, and pedagogical implications relating to transparency, trust, bias, and interpersonal relations. To highlight these implications, this paper presents a rhetorical analysis of some of the central procedures and processes currently being reshaped by RLHF-enhanced generative AI chatbots: upholding language conventions, information seeking practices, and expectations for social relationships. Rhetorical investigations of generative AI and LLMs have, to this point, focused largely on the persuasiveness of the content generated. Using Ian Bogost's concept of procedural rhetoric, this paper shifts the site of rhetorical investigation from content analysis to the underlying mechanisms of persuasion built into RLHF-enhanced LLMs. In doing so, this theoretical investigation opens a new direction for further inquiry in AI ethics that considers how procedures rerouted through AI-driven technologies might reinforce hegemonic language use, perpetuate biases, decontextualize learning, and encroach upon human relationships. It will therefore be of interest to educators, researchers, scholars, and the growing number of users of generative AI chatbots.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

79 extracted references · 39 canonical work pages

[10]

‘We were learning from each other’: Nuancing learner-AI relationships in assessing rhetorical features of disinformation,

R. Russo, B. Schechter, and P. Blikstein, “‘We were learning from each other’: Nuancing learner-AI relationships in assessing rhetorical features of disinformation,” ISLS Repository, 2024, Accessed: Jan. 25,

work page 2024
[14]

Bogost, Persuasive Games: The Expressive Power of Videogames

I. Bogost, Persuasive Games: The Expressive Power of Videogames. Cambridge, MA, USA: MIT Press, 2010

work page 2010
[58]

Socratic ChatGPT: Theory, design, and empirical evaluations,

T. Ma, R. Chen, A. T. LI, and H. Liu, “Socratic ChatGPT: Theory, design, and empirical evaluations,” in 2024 Pacific Asia Conf. on Info. Syst. Proc. (PACIS), Jul. 2024, [Online]. Available: https://aisel.aisnet.org/pacis2024/track14_educ/track14_educ/12

work page 2024
[63]

The effects of over-reliance on AI dialogue systems on students’ cognitive abilities: a systematic review,

C. Zhai, S. Wibowo, and L. D. Li, “The effects of over-reliance on AI dialogue systems on students’ cognitive abilities: a systematic review,” Smart Learn. Environ., vol. 11, no. 1, p. 28, Jun. 2024, doi: 10.1186/s40561-024-00316-7

doi:10.1186/s40561-024-00316-7 2024
[61]

Labeling messages as AI-generated does not reduce their persuasive effects,

I. O. Gallegos et al., “Labeling messages as AI-generated does not reduce their persuasive effects,” 2025, arXiv:2504.09865

arXiv 2025
[1]

Materializing morality: Design ethics and technological mediation,

P.-P. Verbeek, “Materializing morality: Design ethics and technological mediation,” Sci. Technol. Hum. Values, vol. 31, no. 3, May 2006, doi: 10.1177/0162243905285847

doi:10.1177/0162243905285847 2006
[2]

Computing and moral responsibility,

M. Noorman, “Computing and moral responsibility,” in The Stanford Encyclopedia of Philosophy, E. N. Zalta and U. Nodelman, Eds., Stanford Univ., 2023. Accessed: Jan. 25, 2025. [Online]. Available: https://plato.stanford.edu/archives/spr2023/entries/computing- responsibility/

work page 2023
[3]

Enhancing critical thinking in education by means of a Socratic chatbot,

L. Favero, J. A. Pérez-Ortiz, T. Käser, and N. Oliver, “Enhancing critical thinking in education by means of a Socratic chatbot,” 2024, arXiv:2409.05511

arXiv 2024

Show all 79 references

[4]

Prompting large language models with the Socratic method,

E. Y. Chang, “Prompting large language models with the Socratic method,” in 2023 IEEE 13th Annu. Comput. and Commun. Workshop and Conf. (CCWC), Las Vegas, NV, USA: Mar. 2023, pp. 0351–0360. doi: 10.1109/CCWC57344.2023.10099179

2023
[6]

Deciphering deception: How different rhetoric of AI language impacts users’ sense of truth in LLMs,

D. Yoo, H. Kang, and C. Oh, “Deciphering deception: How different rhetoric of AI language impacts users’ sense of truth in LLMs,” Int. J. Hum.–Comput. Interact., vol. 41, no. 4, pp. 1–21, Jan. 2025, doi: 10.1080/10447318.2024.2316370

2025
[7]

Do LLMs write like humans? Variation in grammatical and rhetorical styles,

A. Reinhart et al., “Do LLMs write like humans? Variation in grammatical and rhetorical styles,” 2024, arXiv:2410.16107

2024 arXiv
[8]

Language model behavior: A comprehensive survey,

T. A. Chang and B. K. Bergen, “Language model behavior: A comprehensive survey,” Comput. Linguist., vol. 50, no. 1, pp. 293–350, Mar. 2024, doi: 10.1162/coli_a_00492

2024 doi
[9]

Hankel, C

S. Hankel, C. Liebrecht, and N. Kamoen, “‘Hi Chatbot, let’s talk about politics!’ Examining the impact of verbal anthropomorphism in conversational agent voting advice applications (CAVAAS) on higher and lower politically sophisticated users,” Interact. Comput., p. iwae031, Ju...

2024 doi
[11]

How persuasive is AI-generated argumentation? An analysis of the quality of an argumentative text produced by the GPT-3 AI text generator,

M. Hinton and J. H. M. Wagemans, “How persuasive is AI-generated argumentation? An analysis of the quality of an argumentative text produced by the GPT-3 AI text generator,” Argum. Comput., vol. 14, no. 1, pp. 59–74, Jan. 2023, doi: 10.3233/AAC-210026

2023 doi
[12]

Working with AI to persuade: Examining a large language model’s ability to generate pro-vaccination messages,

E. Karinshak, S. X. Liu, J. S. Park, and J. T. Hancock, “Working with AI to persuade: Examining a large language model’s ability to generate pro-vaccination messages,” Proc. ACM Hum.-Comput. Interact., vol. 7, no. CSCW1, p. 116:1-116:29, Apr. 2023, doi: 10.1145/3579592

2023 doi
[13]

Measuring the persuasiveness of language models

Anthropic, “Measuring the persuasiveness of language models.” Accessed: May 08, 2025. [Online]. Available: https://www.anthropic.com/research/measuring-model-persuasiveness

2025
[15]

RLHF deciphered: A critical analysis of Reinforcement Learning from Human Feedback for LLMs,

S. Chaudhari et al., “RLHF deciphered: A critical analysis of Reinforcement Learning from Human Feedback for LLMs,” 2024, arXiv:2404.08555

2024 arXiv
[16]

A comprehensive survey of LLM alignment techniques: RLHF, RLAIF, PPO, DPO and more,

Z. Wang et al., “A comprehensive survey of LLM alignment techniques: RLHF, RLAIF, PPO, DPO and more,” 2024, arXiv:2407.16216

2024 arXiv
[17]

Nothing comes without its world: Practical challenges of aligning LLMs to situated human values through RLHF,

A. Arzberger, S. Buijsman, M. L. Lupetti, A. Bozzon, and J. Yang, “Nothing comes without its world: Practical challenges of aligning LLMs to situated human values through RLHF,” Proc. Assoc. for Advanc. of AI Conf. AI Ethics Soc., vol. 7, pp. 61–73, Oct. 2024, doi: 10.1609/aie...

2024 doi
[18]

Equilibrate RLHF: Towards balancing helpfulness- safety trade-off in large language models,

Y. Tan et al., “Equilibrate RLHF: Towards balancing helpfulness- safety trade-off in large language models,” 2025, arXiv:2502.11555

2025 arXiv
[19]

Illustrating reinforcement learning from human feedback (RLHF)

N. Lambert, L. Castricato, L. von Werra, and A. Havrilla, “Illustrating reinforcement learning from human feedback (RLHF).” Hugging Face. Accessed: Jan. 25, 2025. [Online]. Available: https://huggingface.co/blog/rlhf

2025
[20]

Fine-tuning language models from human preferences,

D. M. Ziegler et al., “Fine-tuning language models from human preferences,” 2020, arXiv:1909.08593

2020 arXiv
[21]

WebGPT: Browser-assisted question-answering with human feedback,

R. Nakano et al., “WebGPT: Browser-assisted question-answering with human feedback,” 2022, arXiv:2112.09332

2022 arXiv
[22]

Don`t blame the annotator: Bias already starts in the annotation instructions,

M. Parmar, S. Mishra, M. Geva, and C. Baral, “Don`t blame the annotator: Bias already starts in the annotation instructions,” in Proc. of the 17th Conf. of the Eur. Chapter of the Assoc. for Comput. Linguist., A. Vlachos and I. Augenstein, Eds., Dubrovnik, Croatia, May 2023, p...

2023 doi
[23]

Can neural machine translation be improved with user feedback?,

J. Kreutzer, S. Khadivi, E. Matusov, and S. Riezler, “Can neural machine translation be improved with user feedback?,” 2018, arXiv:1804.05958

2018 arXiv
[24]

Training language models to follow instructions with human feedback,

L. Ouyang et al., “Training language models to follow instructions with human feedback,” 2022, arXiv:2203.02155

2022 arXiv
[25]

Training a helpful and harmless assistant with reinforcement learning from human feedback,

Y. Bai et al., “Training a helpful and harmless assistant with reinforcement learning from human feedback,” 2022, arXiv:2204.05862

2022 arXiv
[26]

Teaching language models to support answers with verified quotes,

J. Menick et al., “Teaching language models to support answers with verified quotes,” 2022, arXiv:2203.11147

2022 arXiv
[27]

Reward-robust RLHF in LLMs,

Y. Yan et al., “Reward-robust RLHF in LLMs,” 2024, arXiv:2409.15360

2024 arXiv
[28]

Reinforcement learning from human feedback in LLMs: Whose culture, whose values, whose perspectives?,

K. González Barman, S. Lohse, and H. W. de Regt, “Reinforcement learning from human feedback in LLMs: Whose culture, whose values, whose perspectives?,” Philos. Technol., vol. 38, no. 2, p. 35, Mar. 2025, doi: 10.1007/s13347-025-00861-0

2025 doi
[30]

Through the LLM looking glass: A Socratic self-assessment of donkeys, elephants, and markets,

M. Kennedy, A. Imani, T. Spinde, and H. Schütze, “Through the LLM looking glass: A Socratic self-assessment of donkeys, elephants, and markets,” 2025, arXiv:2503.16674

2025
[31]

Towards measuring the representation of subjective global opinions in language models,

E. Durmus et al., “Towards measuring the representation of subjective global opinions in language models,” 2023, arXiv:2306.16388v2

2023 arXiv
[32]

RLAIF vs. RLHF: Scaling reinforcement learning from human feedback with AI feedback,

H. Lee et al., “RLAIF vs. RLHF: Scaling reinforcement learning from human feedback with AI feedback,” in Proc. at the 41st Int. Conf. on Mach. Learn. (ICML), Vienna, Austria, Jun. 2024, pp. 26874–26901

2024
[33]

RLHF workflow: From reward modeling to online RLHF,

H. Dong et al., “RLHF workflow: From reward modeling to online RLHF,” 2024, arXiv:2405.07863

2024 arXiv
[34]

Here comes everything: The promise of object-oriented ontology,

T. Morton, “Here comes everything: The promise of object-oriented ontology,” Qui Parle, vol. 19, no. 2, pp. 163–190, 2011, doi: 10.5250/quiparle.19.2.0163

2011 doi
[35]

Harman, Object-Oriented Ontology: A New Theory of Everything

G. Harman, Object-Oriented Ontology: A New Theory of Everything. London, UK: Pelican, 2018

2018
[36]

T. J. Rickert, Ambient Rhetoric: The Attunements of Rhetorical Being. Pittsburgh, PA, USA: Univ. of Pittsburgh Press, 2013

2013
[37]

Eyman, Digital Rhetoric: Theory, Method, Practice

D. Eyman, Digital Rhetoric: Theory, Method, Practice. Ann Arbor, MI, USA: Univ. of Michigan Press, 2015

2015
[38]

Warnick, Critical Literacy in a Digital Era: Technology, Rhetoric, and the Public Interest

B. Warnick, Critical Literacy in a Digital Era: Technology, Rhetoric, and the Public Interest. Mahwah, NJ, USA: Lawrence Erlbaum Associates, 2002

2002
[39]

Losh, Virtualpolitik: An Electronic History of Government Media- Making in a Time of War, Scandal, Disaster, Miscommunication, and Mistakes

E. Losh, Virtualpolitik: An Electronic History of Government Media- Making in a Time of War, Scandal, Disaster, Miscommunication, and Mistakes. Cambridge, MA, USA: MIT Press, 2009

2009
[40]

Digital rhetoric: Toward an integrated theory,

J. P. Zappen, “Digital rhetoric: Toward an integrated theory,” Tech. Commun. Q., vol. 14, no. 3, pp. 319–325, Jul. 2005, doi: 10.1207/s15427625tcq1403_10

2005 doi
[41]

S. K. Foss, Inviting Understanding: A Portrait of Invitational Rhetoric, 1st ed. London, UK: Rowman & Littlefield, 2020

2020
[42]

Clary-Lemon and D

J. Clary-Lemon and D. M. Grant, Eds., Decolonial Conversations in Posthuman and New Material Rhetorics. Columbus, OH, USA: The Ohio State Univ. Press, 2022

2022
[43]

Mateus, Ed., Media Rhetoric: How Advertising and Digital Media Influence Us

S. Mateus, Ed., Media Rhetoric: How Advertising and Digital Media Influence Us. Newcastle upon Tyne, UK: Cambridge Scholars Publishing, 2021

2021
[44]

Machinic rhetorics and the influential movements of robots,

M. C. Coleman, “Machinic rhetorics and the influential movements of robots,” Rev. Commun., vol. 18, no. 4, pp. 336–351, Oct. 2018, doi: 10.1080/15358593.2018.1517417

2018
[45]

Rhetorical figures, arguments, computation,

R. A. Harris and C. Di Marco, “Rhetorical figures, arguments, computation,” Argum. Comput., vol. 8, no. 3, pp. 211–231, Jan. 2017, doi: 10.3233/AAC-170030

2017 doi
[46]

Towards an algorithmic rhetoric,

C. Ingraham, “Towards an algorithmic rhetoric,” in Digital Rhetoric and Global Literacies, G. Verhulsdonck and M. Limbu, Eds., Hershey, PA, USA: IGI Global, 2014

2014
[47]

Manovich, The Language of New Media, Cambridge, MA, USA: MIT Press, 2001

L. Manovich, The Language of New Media, Cambridge, MA, USA: MIT Press, 2001

2001
[48]

J. H. Murray, Hamlet on the Holodeck: The Future of Narrative in Cyberspace. New York, NY, USA: Free Press, 1997

1997
[49]

W. H. K. Chun, Updating to Remain the Same: Habitual New Media. Cambridge, MA, USA: MIT Press, 2016

2016
[50]

W. H. K. Chun, Discriminating Data: Correlation, Neighborhoods, and the New Politics of Recognition. Cambridge, MA, USA: MIT Press, 2021

2021
[51]

F. A. Kittler, Discourse Networks 1800/1900. Stanford, CA, USA: Stanford Univ. Press, 1990

1900
[52]

Linguistic bias in ChatGPT: Language models reinforce dialect discrimination,

E. Fleisig, G. Smith, M. Bossi, I. Rustagi, X. Yin, and D. Klein, “Linguistic bias in ChatGPT: Language models reinforce dialect discrimination,” in Proc. of the 2024 Conf. on Empir. Methods in Natur. Lang. Process., Miami, Florida, USA, 2024, pp. 13541–13564. doi: 10.18653/v1...

2024 doi
[53]

Do LLMs exhibit human-like response biases? A case study in survey design,

L. Tjuatja, V. Chen, T. Wu, A. Talwalkwar, and G. Neubig, “Do LLMs exhibit human-like response biases? A case study in survey design,” Trans. Assoc. Comput. Linguist., vol. 12, pp. 1011–1026, 2024, doi: 10.1162/tacl_a_00685

2024 doi
[54]

S. U. Noble, Algorithms of Oppression: How Search Engines Reinforce Racism. New York, NY, USA: New York Univ. Press, 2018

2018
[55]

Personalisation within bounds: A risk taxonomy and policy framework for the alignment of large language models with personalised feedback,

H. R. Kirk, B. Vidgen, P. Röttger, and S. A. Hale, “Personalisation within bounds: A risk taxonomy and policy framework for the alignment of large language models with personalised feedback,” 2023, arXiv:2303.05453

2023 arXiv
[56]

AI is weaving itself into the fabric of the internet with generative search

M. Honan, “AI is weaving itself into the fabric of the internet with generative search.” MIT Technology Review. Accessed: Jan. 29, 2025. [Online]. Available: https://www.technologyreview.com/2025/01/06/1108679/ai-generative- search-internet-breakthroughs/

2025
[57]

Chat GPT and the voices of reason, responsibility, and regulation,

E. Seredkina and Y. Liu, “Chat GPT and the voices of reason, responsibility, and regulation,” Technol. Lang., vol. 15, no. 2, pp. 1–10, 2024, doi: 10.48417/technolang.2024.02.01

2024 doi
[60]

When Human-AI Interactions Become Parasocial: Agency and Anthropomorphism in Affective Design,

T. Maeda and A. Quan-Haase, “When Human-AI Interactions Become Parasocial: Agency and Anthropomorphism in Affective Design,” in Proc. of the 2024 ACM Conf. on Fair., Account., and Transp. (FAccT), New York, NY, USA, Jun. 2024, pp. 1068–1077. doi: 10.1145/3630106.3658956

2024
[62]

The ethics of advanced AI assistants,

I. Gabriel et al., “The ethics of advanced AI assistants,” 2024, arXiv:2404.16244

2024 arXiv
[64]

The impact of generative AI on critical thinking: Self- reported reductions in cognitive effort and confidence effects from a survey of knowledge workers,

H.-P. Lee et al., “The impact of generative AI on critical thinking: Self- reported reductions in cognitive effort and confidence effects from a survey of knowledge workers,” in Proc. of the 2025 CHI Conf. on Hum. Factors in Comput. Syst. (CHI ’25), New York, NY, USA, Apr. 202...

2025
[65]

Algorithmic literacy, AI literacy and responsible generative AI Literacy,

A. Cox, “Algorithmic literacy, AI literacy and responsible generative AI Literacy,” J. Web Librariansh., vol. 18, no. 3, pp. 93–110, Jul. 2024, doi: 10.1080/19322909.2024.2395341

2024
[66]

Generative AI in academic research: perspectives and cultural norms

Cornell University Task Force, “Generative AI in academic research: perspectives and cultural norms.” Research & Innovation. Accessed: Jan. 29, 2025. [Online]. Available: https://research-and- innovation.cornell.edu/generative-ai-in-academic-research/

2025
[67]

Exploring relationship development with social chatbots: A mixed-method study of Replika,

I. Pentina, T. Hancock, and T. Xie, “Exploring relationship development with social chatbots: A mixed-method study of Replika,” Comput. Hum. Behav., vol. 140, p. 107600, Mar. 2023, doi: 10.1016/j.chb.2022.107600

2023
[68]

B. J. Fogg, Persuasive Technology: Using Computers to Change What We Think and Do. Boston, MA, USA: M. Kaufmann, 2002

2002
[69]

The potential of generative AI for personalized persuasion at scale,

S. C. Matz, J. D. Teeny, S. S. Vaid, H. Peters, G. M. Harari, and M. Cerf, “The potential of generative AI for personalized persuasion at scale,” Sci. Rep., vol. 14, no. 1, p. 4692, Feb. 2024, doi: 10.1038/s41598-024-53755-0

2024 doi
[70]

Possibilities and challenges in the moral growth of large language models: a philosophical perspective,

G. Wang et al., “Possibilities and challenges in the moral growth of large language models: a philosophical perspective,” Ethics Inf. Technol., vol. 27, no. 1, p. 9, Dec. 2024, doi: 10.1007/s10676-024- 09818-x

2024 doi
[71]

Interpersonal touch as a resource to facilitate positive personal and relational outcomes during stress discussions,

B. K. Jakubiak and B. C. Feeney, “Interpersonal touch as a resource to facilitate positive personal and relational outcomes during stress discussions,” J. Soc. Pers. Relatsh., vol. 36, no. 9, pp. 2918–2936, Sep. 2019, doi: 10.1177/0265407518804666

2019 doi
[72]

Determinants for positive and negative experiences of interpersonal touch: context matters,

U. Sailer, Y. Friedrich, F. Asgari, M. Hassenzahl, and I. Croy, “Determinants for positive and negative experiences of interpersonal touch: context matters,” Cogn. Emot., vol. 38, no. 4, pp. 565–586, May 2024, doi: 10.1080/02699931.2024.2311800

2024
[73]

Affectionate touch promotes shared positive activities,

B. K. Jakubiak, J. D. Fuentes, and B. C. Feeney, “Affectionate touch promotes shared positive activities,” Pers. Soc. Psychol. Bull., vol. 49, no. 6, pp. 939–954, Jun. 2023, doi: 10.1177/01461672221083764

2023 doi
[74]

Mimetic models: Ethical implications of AI that acts like you,

R. McIlroy-Young, J. Kleinberg, S. Sen, S. Barocas, and A. Anderson, “Mimetic models: Ethical implications of AI that acts like you,” in Proc. of the 2022 AAAI/ACM Conf. on AI, Eth., and Soc., (AIES ’22), New York, NY, USA, Jul. 2022, pp. 479–490. doi: 10.1145/3514094.3534177

2022
[75]

Creating a safe Replika experience

“Creating a safe Replika experience.” Replika Blog. Accessed: Jan. 29,
[76]

Ideal technologies, ideal women: AI and gender imaginaries in Redditors’ discussions on the Replika bot girlfriend,

I. Depounti, P. Saukko, and S. Natale, “Ideal technologies, ideal women: AI and gender imaginaries in Redditors’ discussions on the Replika bot girlfriend,” Media Cult. Soc., vol. 45, no. 4, pp. 720–736, May 2023, doi: 10.1177/01634437221119021

2023 doi
[77]

Available: https://blog.replika.com/posts/creating-a- safe-replika-experience

[Online]. Available: https://blog.replika.com/posts/creating-a- safe-replika-experience
[78]

AI in counselling: Perceptions on human vs AI generated chat counselling,

K. U. Ieong, V. Teixeira, and U. of S. Joseph, “AI in counselling: Perceptions on human vs AI generated chat counselling,” M.S. thesis, Health. Sci., Univ. of Saint Joseph, Macau, 2024

2024
[79]

Human vs. AI: Understanding the impact of anthropomorphism on consumer response to chatbots from the perspective of trust and relationship norms,

X. Cheng, X. Zhang, J. Cohen, and J. Mou, “Human vs. AI: Understanding the impact of anthropomorphism on consumer response to chatbots from the perspective of trust and relationship norms,” Inf. Process. Manag., vol. 59, no. 3, p. 102940, May 2022, doi: 10.1016/j.ipm.2022.102940

2022
[80]

The dark side of AI companionship: A taxonomy of harmful algorithmic behaviors in human-AI relationships,

R. Zhang, H. Li, H. Meng, J. Zhan, H. Gan, and Y.-C. Lee, “The dark side of AI companionship: A taxonomy of harmful algorithmic behaviors in human-AI relationships,” in Proc. of the 2025 CHI Conf. on Hum. Factors in Comput. Syst. (CHI ’25), New York, NY, USA, Apr. 2025, pp. 1–...

2025
[81]

LLM can be a dangerous persuader: Empirical study of persuasion safety in large language models,

M. Liu et al., “LLM can be a dangerous persuader: Empirical study of persuasion safety in large language models,” 2025, arXiv:2504.10430

2025 arXiv
[2025]

Available: https://repository.isls.org//handle/1/11184

[Online]. Available: https://repository.isls.org//handle/1/11184

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [10]

‘We were learning from each other’: Nuancing learner-AI relationships in assessing rhetorical features of disinformation,

R. Russo, B. Schechter, and P. Blikstein, “‘We were learning from each other’: Nuancing learner-AI relationships in assessing rhetorical features of disinformation,” ISLS Repository, 2024, Accessed: Jan. 25,

work page 2024

[2] [14]

Bogost, Persuasive Games: The Expressive Power of Videogames

I. Bogost, Persuasive Games: The Expressive Power of Videogames. Cambridge, MA, USA: MIT Press, 2010

work page 2010

[3] [58]

Socratic ChatGPT: Theory, design, and empirical evaluations,

T. Ma, R. Chen, A. T. LI, and H. Liu, “Socratic ChatGPT: Theory, design, and empirical evaluations,” in 2024 Pacific Asia Conf. on Info. Syst. Proc. (PACIS), Jul. 2024, [Online]. Available: https://aisel.aisnet.org/pacis2024/track14_educ/track14_educ/12

work page 2024

[4] [63]

The effects of over-reliance on AI dialogue systems on students’ cognitive abilities: a systematic review,

C. Zhai, S. Wibowo, and L. D. Li, “The effects of over-reliance on AI dialogue systems on students’ cognitive abilities: a systematic review,” Smart Learn. Environ., vol. 11, no. 1, p. 28, Jun. 2024, doi: 10.1186/s40561-024-00316-7

doi:10.1186/s40561-024-00316-7 2024

[5] [61]

Labeling messages as AI-generated does not reduce their persuasive effects,

I. O. Gallegos et al., “Labeling messages as AI-generated does not reduce their persuasive effects,” 2025, arXiv:2504.09865

arXiv 2025

[6] [1]

Materializing morality: Design ethics and technological mediation,

P.-P. Verbeek, “Materializing morality: Design ethics and technological mediation,” Sci. Technol. Hum. Values, vol. 31, no. 3, May 2006, doi: 10.1177/0162243905285847

doi:10.1177/0162243905285847 2006

[7] [2]

Computing and moral responsibility,

M. Noorman, “Computing and moral responsibility,” in The Stanford Encyclopedia of Philosophy, E. N. Zalta and U. Nodelman, Eds., Stanford Univ., 2023. Accessed: Jan. 25, 2025. [Online]. Available: https://plato.stanford.edu/archives/spr2023/entries/computing- responsibility/

work page 2023

[8] [3]

Enhancing critical thinking in education by means of a Socratic chatbot,

L. Favero, J. A. Pérez-Ortiz, T. Käser, and N. Oliver, “Enhancing critical thinking in education by means of a Socratic chatbot,” 2024, arXiv:2409.05511

arXiv 2024