pith. sign in

arxiv: 2606.19864 · v1 · pith:EB3CPH2Cnew · submitted 2026-06-18 · 💻 cs.CL

The Almost Intelligent Revolution: Options for Scaling Up Deliberation and Empowering People with AI

Serge Sharoff This is my paper

Pith reviewed 2026-06-26 17:41 UTC · model grok-4.3

classification 💻 cs.CL
keywords Large Language Modelsdemocratic deliberationinclusivitySystemic-Functional Linguisticsargumentation scaffoldinglinguistic biasesmarginalized groupsAI ethics
0
0 comments X

The pith

Large language models can scale up and democratize deliberation by scaffolding argumentation and reducing linguistic biases that exclude marginalized groups.

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

The paper argues that LLMs offer a practical way to expand public deliberation and make it more inclusive for people from varied backgrounds. It examines how differences in language use tied to socio-demographic groups and communicative purposes affect who joins in and how effectively. Using concepts from Systemic-Functional Linguistics, the work shows AI can structure arguments, improve access, and lessen the pull of formal language styles that often sideline certain voices. A sympathetic reader would care because this points toward more representative decision-making processes if the technology is guided properly. The chapter balances this with calls for ethical safeguards to avoid reinforcing existing inequalities.

Core claim

LLMs can be used to significantly scale up and democratise deliberation, particularly in fostering inclusivity and empowering traditionally marginalised groups by scaffolding argumentation, enhancing access, and reducing the influence of exclusionary linguistic norms and biases embedded in prestigious registers.

What carries the argument

Systemic-Functional Linguistics analysis of variations across language users and language use applied to AI-supported deliberation systems.

If this is right

  • AI can scaffold argumentation structures for participants from different socio-demographic groups.
  • Access to deliberation processes improves when AI accounts for varied communicative functions.
  • The influence of exclusionary norms in prestigious registers decreases through targeted AI support.
  • Ethical safeguards embedded in the systems can help prevent reproduction of linguistic inequalities.

Where Pith is reading between the lines

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

  • Deliberation platforms could incorporate real-time language adaptation features to match individual user patterns.
  • The approach might extend to other civic AI uses such as policy feedback collection.
  • Comparative studies of participation rates with and without scaffolding would provide direct tests.
  • Integration into existing social media or forum tools could expand reach to new user segments.

Load-bearing premise

Variations across socio-demographic groups and communicative functions can be addressed by AI to improve participation without reproducing linguistic inequalities.

What would settle it

A controlled experiment on deliberation platforms that finds LLM assistance increases rather than decreases the dominance of prestigious language registers among participants from marginalized groups.

Figures

Figures reproduced from arXiv: 2606.19864 by Serge Sharoff.

Figure 1
Figure 1. Figure 1: Linguistic interaction in society, following Halliday (1999) [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
read the original abstract

The increasing prominence of Large Language Models (LLMs) in public discourse presents both opportunities and challenges for democratic deliberation. While red teaming strategies help mitigate specific risks, broader concerns persist regarding linguistic constraints, biases, and the sycophantic tendencies of LLMs. This chapter explores how LLMs can be used to significantly scale up and democratise deliberation, particularly in fostering inclusivity and empowering traditionally marginalised groups. Drawing on concepts from Systemic-Functional Linguistics, the chapter examines how variations across language users (for example, with respect to socio-demographic groups) and across language use (for example, with respect to communicative functions) shape participation in AI-supported deliberation. The chapter presents AI-driven deliberation studies and assesses their potential to scaffold argumentation, enhance access, and reduce the influence of exclusionary linguistic norms and biases which are embedded in prestigious registers. At the same time, the chapter cautions against both overclaiming, which leads to unrealistic expectations, and underclaiming, which risks missed opportunities for AI-assisted engagement. The chapter concludes by identifying future research directions to maximise the democratic potential of AI-assisted participation while embedding ethical safeguards to counteract the reproduction of linguistic inequalities.

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.

Referee Report

0 major / 2 minor

Summary. The paper claims that LLMs offer opportunities to scale up and democratize democratic deliberation by fostering inclusivity and empowering marginalized groups. It draws on Systemic-Functional Linguistics to examine how socio-demographic variations in language users and communicative functions in language use shape AI-supported participation. The chapter reviews existing AI-driven deliberation studies, assesses their potential to scaffold argumentation, enhance access, and mitigate exclusionary linguistic norms and biases in prestigious registers, while cautioning against overclaiming and underclaiming, and identifies future research directions with embedded ethical safeguards.

Significance. If the conceptual synthesis holds, the work contributes by bridging Systemic-Functional Linguistics with AI deliberation applications, providing a framework for identifying risks of linguistic inequality reproduction and outlining conditional pathways for more inclusive AI-assisted civic engagement.

minor comments (2)
  1. [Abstract] Abstract: The claim that the chapter 'presents AI-driven deliberation studies' would benefit from explicit citation of the specific studies reviewed and a brief indication of their key findings or limitations to ground the assessment of potential.
  2. The manuscript positions itself as exploratory and refers to external studies without new empirical data; ensuring that all referenced studies are cited with sufficient detail in the main text would strengthen readability for an interdisciplinary audience.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive assessment of the manuscript, the accurate summary of its contributions, and the recommendation for minor revision. The referee's evaluation correctly identifies the paper's focus on bridging Systemic-Functional Linguistics with AI-supported deliberation while maintaining appropriate cautions.

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper is an exploratory conceptual review that surveys Systemic-Functional Linguistics concepts and existing AI-deliberation literature without advancing any formal model, equations, fitted parameters, or primary empirical claims. No derivation chain exists that could reduce to self-definition, fitted inputs renamed as predictions, or load-bearing self-citations; all references to external studies and domain concepts remain independent of the present text. The strongest claims are explicitly conditional on future research and ethical safeguards rather than internally forced by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The central claim rests on domain assumptions from linguistics and democratic theory about the addressability of language biases by AI, without new foundational elements or independent evidence provided in the abstract.

axioms (1)
  • domain assumption Systemic-Functional Linguistics provides a useful framework for analyzing how language variations across users and functions affect AI-supported deliberation.
    Invoked explicitly in the abstract to examine variations and their impact on participation.

pith-pipeline@v0.9.1-grok · 5729 in / 1250 out tokens · 42445 ms · 2026-06-26T17:41:13.792715+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

35 extracted references · 4 canonical work pages · 2 internal anchors

  1. [1]

    Oxford University Press, 2018

    André Bächtiger, John S Dryzek, Jane Mansbridge, and Mark E Warren.The Oxford handbook of deliberative democracy. Oxford University Press, 2018

    2018

  2. [2]

    Michael A. K. Halliday and Christian M. I. M. Matthiessen.Construing experience through meaning: a language-based approach to cognition. Cassell, London, 1999

    1999

  3. [3]

    GPT-4 passes the bar exam.Philosophical Transactions of the Royal Society A, 382(2270), 2024

    Daniel Martin Katz, Michael James Bommarito, Shang Gao, and Pablo Arredondo. GPT-4 passes the bar exam.Philosophical Transactions of the Royal Society A, 382(2270), 2024

    2024

  4. [4]

    Michael A. K. Halliday. The notion of “context” in language education. In Mohsen Ghadessy, editor, Text and Context in Functional Linguistics, pages 1–24. John Benjamins, Amsterdam, 1999

    1999

  5. [5]

    Register in the round: registerial cartography.Functional Linguistics, 2(1):1–48, 2015

    Christian MIM Matthiessen. Register in the round: registerial cartography.Functional Linguistics, 2(1):1–48, 2015

    2015

  6. [6]

    Genre annotation for the web: text-external and text-internal perspectives.Register studies, 3:1–32, 2021

    Serge Sharoff. Genre annotation for the web: text-external and text-internal perspectives.Register studies, 3:1–32, 2021

    2021

  7. [7]

    Inclusion Europe, Brussels, Belgium, 2009

    Inclusion Europe.Information for All: European Guidelines for the Production of Easy-to-Read Infor- mation. Inclusion Europe, Brussels, Belgium, 2009

    2009

  8. [8]

    Language models are few-shot learners.Advances in neural information processing systems, 33:1877–1901, 2020

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, Sandhini Agarwal, Ariel Herbert-V oss, Gretchen Krueger, Tom Henighan, Rewon Child, Aditya Ramesh, Daniel Ziegler, Jeffrey Wu, Clemens Winter, Chris Hesse, Mark Chen, Eric Sigler, Mateusz Litwin, Scott G...

    1901

  9. [9]

    Form and function: automatic methods for prediction of functions

    Serge Sharoff. Form and function: automatic methods for prediction of functions. In Rebekah We- gener, Anne McCabe, Akila Sellami-Baklouti, and Lise Fontaine, editors,Transdisciplinary Systemic Functional Linguistics. Routledge, 2025

    2025

  10. [10]

    Kevin Roose. A very strange conversation with the chatbot built into Microsoft’s search engine led to it declaring its love for me.New York Times, Section A:1, Feb 2023.https://www.nytimes.com/ 2023/02/16/technology/bing-chatbot-microsoft-chatgpt.html

    2023

  11. [11]

    Artificial intelligence, values, and alignment.Minds and machines, 30(3):411–437, 2020

    Iason Gabriel. Artificial intelligence, values, and alignment.Minds and machines, 30(3):411–437, 2020

    2020

  12. [12]

    Zorik Gekhman, Gal Yona, Roee Aharoni, Matan Eyal, Amir Feder, Roi Reichart, and Jonathan Herzig. Does fine-tuning LLMs on new knowledge encourage hallucinations? InProceedings of the 2024 Conference on Empirical Methods in Natural Language Processing, pages 7765–7784, Miami, Florida, USA, November 2024. Association for Computational Linguistics

    2024

  13. [13]

    AI can help humans find common ground in democratic deliberation.Science, 386(6719), 2024

    Michael Henry Tessler, Michiel A Bakker, Daniel Jarrett, Hannah Sheahan, Martin J Chadwick, Raphael Koster, Georgina Evans, Lucy Campbell-Gillingham, Tantum Collins, David C Parkes, et al. AI can help humans find common ground in democratic deliberation.Science, 386(6719), 2024

    2024

  14. [14]

    Artificial intelligence in deliberation: The ai penalty and the emergence of a new deliberative divide.arXiv preprint arXiv:2503.07690, 2025

    Andreas Jungherr and Adrian Rauchfleisch. Artificial intelligence in deliberation: The ai penalty and the emergence of a new deliberative divide.arXiv preprint arXiv:2503.07690, 2025

    work page arXiv 2025

  15. [15]

    Democracy made easy: Simplifying complex topics to enable democratic participation

    Nouran Khallaf, Stefan Bott, Carlo Eugeni, John O’Flaherty, Serge Sharoff, and Horacio Saggion. Democracy made easy: Simplifying complex topics to enable democratic participation. InProc Artifi- cial Intelligence and Easy and Plain Language in Institutional Context at Machine Translation Summit, Geneva, June 2025

    2025

  16. [16]

    Unsupervised cross-lingual representation learning at scale

    Alexis Conneau, Kartikay Khandelwal, Naman Goyal, Vishrav Chaudhary, Guillaume Wenzek, Fran- cisco Guzmán, Edouard Grave, Myle Ott, Luke Zettlemoyer, and Veselin Stoyanov. Unsupervised cross-lingual representation learning at scale. InProceedings of the 58th Annual Meeting of the As- sociation for Computational Linguistics, pages 8440–8451, Online, July 2020

    2020

  17. [17]

    Salamandra technical report, 2025

    Aitor Gonzalez-Agirre, Marc Pàmies, Joan Llop, Irene Baucells, Severino Da Dalt, Daniel Tamayo, José Javier Saiz, Ferran Espuña, Jaume Prats, Javier Aula-Blasco, Mario Mina, Adrián Rubio, Alexander Shvets, Anna Sallés, Iñaki Lacunza, Iñigo Pikabea, Jorge Palomar, Júlia Falcão, Lucía Tormo, Luis Vasquez-Reina, Montserrat Marimon, Valle Ruíz-Fernández, and ...

    2025

  18. [18]

    From her story, to our story: Digital storytelling as public engagement around abortion rights advocacy in Ireland

    Lydia Michie, Madeline Balaam, John McCarthy, Timur Osadchiy, and Kellie Morrissey. From her story, to our story: Digital storytelling as public engagement around abortion rights advocacy in Ireland. In Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems, pages 1–15, 2018

    2018

  19. [19]

    Probing a community-based conversational storytelling agent to document digital stories of housing insecurity

    Brett A Halperin, Gary Hsieh, Erin McElroy, James Pierce, and Daniela K Rosner. Probing a community-based conversational storytelling agent to document digital stories of housing insecurity. InProceedings of the 2023 CHI Conference on Human Factors in Computing Systems, pages 1–18, 2023. 13

    2023

  20. [20]

    Five sources of bias in natural language processing.Language and linguistics compass, 15(8):e12432, 2021

    Dirk Hovy and Shrimai Prabhumoye. Five sources of bias in natural language processing.Language and linguistics compass, 15(8):e12432, 2021

    2021

  21. [21]

    The weirdest people in the world?Behavioral and brain sciences, 33(2-3):61–83, 2010

    Joseph Henrich, Steven J Heine, and Ara Norenzayan. The weirdest people in the world?Behavioral and brain sciences, 33(2-3):61–83, 2010

    2010

  22. [22]

    Learning the difference that makes a difference with counterfactually-augmented data

    Divyansh Kaushik, Eduard Hovy, and Zachary Lipton. Learning the difference that makes a difference with counterfactually-augmented data. InInternational Conference on Learning Representations, 2020

    2020

  23. [23]

    Bowman, Amanda Askell, Roger Grosse, Danny Hernandez, Deep Ganguli, Evan Hubinger, Nicholas Schiefer, and Jared Kaplan

    Ethan Perez, Sam Ringer, Kamile Lukosiute, Karina Nguyen, Edwin Chen, Scott Heiner, Craig Pet- tit, Catherine Olsson, Sandipan Kundu, Saurav Kadavath, Andy Jones, Anna Chen, Benjamin Mann, Brian Israel, Bryan Seethor, Cameron McKinnon, Christopher Olah, Da Yan, Daniela Amodei, Dario Amodei, Dawn Drain, Dustin Li, Eli Tran-Johnson, Guro Khundadze, Jackson ...

    2023

  24. [24]

    Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned

    Deep Ganguli, Liane Lovitt, Jackson Kernion, Amanda Askell, Yuntao Bai, Saurav Kadavath, Ben Mann, Ethan Perez, Nicholas Schiefer, Kamal Ndousse, et al. Red teaming language models to reduce harms: Methods, scaling behaviors, and lessons learned.arXiv preprint arXiv:2209.07858, 2022

    work page internal anchor Pith review Pith/arXiv arXiv 2022

  25. [25]

    Integrating artificial intelligence into citizens’ assemblies: Benefits, concerns and future pathways.Journal of Deliberative Democracy, 20(1), 2024

    Sammy McKinney. Integrating artificial intelligence into citizens’ assemblies: Benefits, concerns and future pathways.Journal of Deliberative Democracy, 20(1), 2024

    2024

  26. [26]

    Large language models (LLMs) as agents for augmented democracy.Philosophical Transactions A, 382(20240100):1–17, 2024

    Jairo F Gudiño, Umberto Grandi, and César Hidalgo. Large language models (LLMs) as agents for augmented democracy.Philosophical Transactions A, 382(20240100):1–17, 2024

    2024

  27. [27]

    The dangers of underclaiming: Reasons for caution when reporting how NLP sys- tems fail

    Samuel Bowman. The dangers of underclaiming: Reasons for caution when reporting how NLP sys- tems fail. InProceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 7484–7499, Dublin, Ireland, May 2022. Association for Computational Linguistics

    2022

  28. [28]

    Achieving Human Parity on Automatic Chinese to English News Translation

    Hany Hassan, Anthony Aue, Chang Chen, Vishal Chowdhary, Jonathan Clark, Christian Federmann, Xuedong Huang, Marcin Junczys-Dowmunt, William Lewis, Mu Li, Shujie Liu, Tie-Yan Liu, Renqian Luo, Arul Menezes, Tao Qin, Frank Seide, Xu Tan, Fei Tian, Lijun Wu, Shuangzhi Wu, Yingce Xia, Dongdong Zhang, Zhirui Zhang, and Ming Zhou. Achieving human parity on auto...

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  29. [29]

    Welcome to the era of experience

    David Silver and Richard S Sutton. Welcome to the era of experience. In George Konidaris, editor, Designing an Intelligence. MIT Press, 2025. 14

    2025

  30. [30]

    Nick Bostrom.Superintelligence: Paths, dangers, strategies.Oxford University Press, 2014

    2014

  31. [31]

    Classification of global catastrophic risks connected with artificial intelligence.AI & Society, 35(1):147–163, 2020

    Alexey Turchin and David Denkenberger. Classification of global catastrophic risks connected with artificial intelligence.AI & Society, 35(1):147–163, 2020

    2020

  32. [32]

    On the dangers of stochastic parrots: Can language models be too big? InProceedings of the 2021 ACM conference on fairness, accountability, and transparency, pages 610–623, 2021

    Emily M Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell. On the dangers of stochastic parrots: Can language models be too big? InProceedings of the 2021 ACM conference on fairness, accountability, and transparency, pages 610–623, 2021

    2021

  33. [33]

    A set of recommendations for assessing human–machine parity in language translation.Journal of Artificial Intelligence Research, 67:653–672, 2020

    Samuel Läubli, Sheila Castilho, Graham Neubig, Rico Sennrich, Qinlan Shen, and Antonio Toral. A set of recommendations for assessing human–machine parity in language translation.Journal of Artificial Intelligence Research, 67:653–672, 2020

    2020

  34. [34]

    To add AI, or not to add AI?Knowledge-Based Systems, 2(2):128–132, 1989

    Derek Partridge. To add AI, or not to add AI?Knowledge-Based Systems, 2(2):128–132, 1989

    1989

  35. [35]

    Neuro-symbolic AI in 2024: A systematic review.arXiv preprint arXiv:2501.05435, 2025

    Brandon C Colelough and William Regli. Neuro-symbolic AI in 2024: A systematic review.arXiv preprint arXiv:2501.05435, 2025. 15

    work page arXiv 2024