The Almost Intelligent Revolution: Options for Scaling Up Deliberation and Empowering People with AI
Pith reviewed 2026-06-26 17:41 UTC · model grok-4.3
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.
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
- 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
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.
Referee Report
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)
- [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.
- 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
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
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
axioms (1)
- domain assumption Systemic-Functional Linguistics provides a useful framework for analyzing how language variations across users and functions affect AI-supported deliberation.
Reference graph
Works this paper leans on
-
[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]
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]
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]
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]
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]
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]
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]
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]
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]
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]
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]
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]
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]
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
-
[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]
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]
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]
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]
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]
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]
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]
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]
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]
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]
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]
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]
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]
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]
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]
Nick Bostrom.Superintelligence: Paths, dangers, strategies.Oxford University Press, 2014
2014
-
[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]
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]
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]
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]
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
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.