Pith. sign in

REVIEW 3 major objections 5 minor 27 references

Emphasizing Deliberation and Critical Thinking in an AI Hype World

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read To counter AI solutionism, the paper argues for deliberate, slow research and critical engagement—including non-engagement—rather than bans.

desk verdict A clear, honest position paper that makes a modest but sensible case for deliberation and critical thinking as a middle path against AI solutionism; the main soft spot is its reliance on the unargued premise that human critical thinking stays ahead of AI. read the letter →

arxiv 2507.14961 v1 pith:OJCNATAD submitted 2025-07-20 cs.HC

Katja Rogers This is my paper
classification cs.HC
keywords AIsolutionismcriticalthinkingdeliberationnon-engagementgenerativehypehuman-computerinteractionslowresearchnovelty
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

The paper argues that resisting AI solutionism should not mean trying to ban or abandon the technology, which is unlikely to work and probably not beneficial. Instead, the author proposes resistance through slower, more deliberate research, conscientious critical engagement with AI outputs, and the legitimate choice of non-engagement. The reason this matters is that early and unthinking reliance on generative AI may stymie critical thinking in students and amplify existing poor research practices. If true, the path forward is not policy prohibition but a change in individual and community norms about when, where, and why to use AI.

What carries the argument

The central mechanism is the 'when, where, and why' questioning heuristic, borrowed from discussions of conference travel, applied to AI use. It turns resistance from a blanket rejection into a case-by-case weighing of trade-offs, and it anchors that weighing in critical thinking as a skill that must be maintained through education and practice.

What would settle it

A controlled longitudinal study in which students who rely on generative AI from early on score equal to or better on critical-thinking assessments than students who are trained to deliberate without it would undercut the claim that early reliance stymies critical thinking.

Watch

Extended reading notes

Core claim

The central claim is that human judgement and critical thinking remain a crucial component in human-AI interaction, and that deliberately exercising them is a form of resistance to AI solutionism. The author contends that banning AI is neither feasible nor desirable, and that the better strategy is conscientious engagement: think before designing or using AI, use saved time for social or quality-improving activities, and redesign education to teach and assess critical thinking rather than generating ChatGPT-gradable output. This stance rests on the belief that AI, while improving, has not mastered and may never master high-level critical thinking, so human deliberation retains durable value.

Load-bearing premise

The argument relies on the assumption that AI systems will not achieve high-level critical thinking skills, so that human deliberation remains a durable advantage and an effective form of resistance.

Editorial extensions

If this is right

  • Bans and abandonment are not the goal; instead, individuals and communities should ask when, where, and why AI is appropriate before use or design.
  • Time saved by AI should be redirected toward social or interpersonal good, or toward improving the quality and meaningfulness of work, not just more output.
  • Courses should be redesigned so learning objectives include and assess critical thinking, for example through exam formats that avoid grading ChatGPT output.
  • Research communities should push back against incentives for quantity and speed and reward quality and future quality improvements.

Reading between the lines

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

  • The author's stance implies that AI literacy education should focus on critical evaluation of AI output rather than tool proficiency alone.
  • A testable extension would be comparing research or educational outcomes when non-engagement is explicitly framed as a valid choice.
  • The argument could also reshape review criteria, for example by requiring authors to state how they deliberated over AI use.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. This 5-page position paper argues that the hype surrounding AI is amplified by HCI's emphasis on novelty, and that resisting 'AI solutionism' via outright bans or abandonment is neither feasible nor desirable. Instead, it advocates slow(er), deliberate use of AI, conscientious critical engagement, and explicit non-engagement, together with redesign of HCI courses and research practices to protect critical thinking and knowledge quality. The paper supports this stance with two worries (students' critical thinking and the entrenchment of poor reporting habits), draws on a small set of recent CHI references and an informal ACM Library search, and closes with concrete personal strategies. The central claim is explicitly hedged as 'may help.'

Significance. For a workshop on resisting AI solutionism, the paper offers a clearly written, constructive middle path between technological solutionism and blanket rejection, and it honestly labels its empirical basis as anecdotal. The explicit hedging, the admission that skepticism about future AI capabilities is a personal belief, and the non-circular use of the author's own prior work [17] are commendable. Its broader value is limited by the absence of systematic evidence for the proposed strategies and by its reliance on one forward-looking factual assumption about human superiority in critical thinking, though the paper's transparency about these limitations makes it a defensible position statement if the requested revisions are made.

major comments (3)
  1. [Section 2] Paragraph beginning 'AI technologies are getting better and better': the prescriptive force of the paper rests on an unargued factual premise that AI will not achieve high-level critical thinking skills. The author writes that they have 'not yet seen it capable' and 'remain skeptical that it will manage this in the future,' but gives no evidence or argument for this durability. If future systems reach or exceed expert-level argument evaluation, hidden-assumption detection, and metacognitive judgment in novel contexts, then the proposed strategy (deliberation, critical engagement, and non-engagement as resistance) loses its distinguishing advantage. The paper should either support this premise with empirical or analytic argument, or explicitly conditionalize the recommendation on it and discuss the implications if the premise fails; re-grounding in values such as accountability or democratic legitimacy, which are only mentioned in passing, would be one way to make the recommendation robust.
  2. [Section 2] Paragraphs from 'Pushing back more firmly...' through the numbered strategy list: the move from the two harms (student critical thinking, knowledge/reporting quality) to the proposed strategies ('think before we use or design AI,' use freed time for quality or social good, redesign courses to assess critical thinking) is not accompanied by an account of the mechanism by which these actions will reduce harmful impacts or build a solid knowledge foundation. Because the abstract's 'may help' is a predictive claim, the manuscript should distinguish value-based commitments from evidence-based interventions and, ideally, identify observable indicators or a research agenda that could test the effectiveness of slow research and non-engagement. Without this, the central recommendation remains an untested opinion, which is acceptable for a position paper only if explicitly framed as such.
  3. [Section 1 and Section 2] The paper acknowledges that AI tools can be genuinely beneficial, e.g., for non-native English speakers or dyslexic researchers (Section 1, 'There are good reasons for excitement'), yet later recommends non-engagement as part of the resistance strategy (Section 2, 'Overall, for me, resisting AI solutionism means...'). The manuscript does not explain how non-engagement can be encouraged without inequitably affecting those who rely on AI for accessibility or language support. Please address this tension, for instance by clarifying that non-engagement is a context-specific individual choice and discussing how community advocacy would handle such dependencies.
minor comments (5)
  1. [Figure 1] Figure 1 is presented without a methods note: provide the date of the search, the searched fields (e.g., title, abstract, or full text), and whether results were deduplicated or normalized; currently the comparison is an anecdote rather than a reproducible query.
  2. [Section 1] The phrase 'AI solutionism' is central to the argument but is not defined; please define it on first use, for example by explicitly engaging with the term as used in the cited workshop [16].
  3. [Section 2] In the enumeration '1) ... 2) ... 3) ...' the formatting and punctuation are inconsistent; use a consistent list style with complete sentences for each item.
  4. [References] Reference [16] lists 'CHI EA ’24' while the paper's context is a CHI ’25 workshop; please verify the correct year and proceedings series.
  5. [Section 1] The informal phrase 'doesn’t need much belabouring' may be better phrased as 'scarcely needs elaboration' to match the register of the rest of the paper.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is an opinion piece without a predictive derivation, and its self-citations are illustrative rather than load-bearing.

full rationale

This is a position/opinion paper arguing for deliberation and critical thinking as resistance to AI solutionism. It makes no predictive or derived quantitative claims, fits no parameters, and presents no formal derivation. The author's self-citations [17] (Rogers 2025) and [18] (Rogers et al. 2024) are used only to illustrate an existing concern about reporting quality (e.g., 'if LLMs suggest content based on the current state of things, then poor reporting habits may be amplified... [17]'); this is a supporting reference, not a load-bearing theorem or fitted input. The paper's central premise that human critical thinking may remain something AI cannot replicate is an explicitly subjective forward-looking belief ('I remain skeptical that it will manage this in the future'), which is an open empirical assumption, not a circular derivation from the paper's own inputs. No equation or definition in the paper reduces to itself, and no prediction is tested against data. Accordingly, no circularity is present; the appropriate score is 0.

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

No fitted parameters or invented entities. The paper relies on two domain assumptions about AI capabilities and the value of human judgment, plus a normative assumption that deliberation is an effective form of resistance.

assumptions (3)
  • domain assumption Human judgment and expertise remain a critical component in human-AI interaction.
    Stated in Section 2: 'I subscribe to the idea that human judgement and expertise remain a critical component in human-AI interaction.' This is a value-laden assumption that underpins the call for critical thinking.
  • domain assumption AI systems will not master high-level critical thinking in the foreseeable future.
    The author's skepticism in Section 2: 'I have not yet seen it capable of mastering or even successfully masquerading what I would consider high-level critical thinking skills, and I remain skeptical that it will manage this in the future.' This is a predictive assumption that justifies why critical thinking remains a human advantage.
  • ad hoc to paper Deliberate, slow use can reduce harmful impacts without banning technology.
    The core recommendation in the conclusion; it is asserted as a belief ('I firmly believe') without empirical support.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Emphasizing Deliberation and Critical Thinking in an AI Hype World." pith.science (2026). https://pith.science/paper/OJCNATAD

@misc{pith2026250714961,
  author       = {Pith},
  title        = {Pith review of: Emphasizing Deliberation and Critical Thinking in an AI Hype World},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/OJCNATAD}},
  note         = {Machine review of arXiv:2507.14961}
}
read the original abstract

AI solutionism is accelerated and substantiated by hype and HCI's elevation of novelty. Banning or abandoning technology is unlikely to work and probably not beneficial on the whole either -- but slow(er), deliberate use together with conscientious, critical engagement and non-engagement may help us navigate a post-AI hype world while contributing to a solid knowledge foundation and reducing harmful impacts in education and research.

Figures

Figures reproduced from arXiv: 2507.14961 by the authors.

Figure 1
Figure 1. Left: 21,470 results for "novel" in ACM library filtered to CHI proceedings. Right: In comparison, searching for "human-computer [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

27 extracted references · 18 canonical work pages

  1. [17]

    The Shiny Scary Future of Automated Research Synthesis in HCI

    Katja Rogers. 2025. The Shiny Scary Future of Automated Research Synthesis in HCI. arXiv:2501.16084 [cs.HC] https://arxiv.org/abs/2501.16084 Accepted at CHI 2024 workshop: "LLMs as Research Tools: Applications and Evaluations in HCI Data Work"

    work page Pith review arXiv 2025

  2. [1]

    Baumer and M

    Eric P.S. Baumer and M. Six Silberman. 2011. When the implication is not to design (technology). In Proceedings of the SIGCHI Conference on Human Factors in Computing Systems (Vancouver, BC, Canada) (CHI ’11). Association for Computing Machinery, New York, NY, USA, 2271–2274. doi:10.1145/1978942.1979275

    arXiv 2011

  3. [2]

    Courtni Byun, Piper Vasicek, and Kevin Seppi. 2023. Dispensing with Humans in Human-Computer Interaction Research. In Extended Abstracts of the 2023 CHI Conference on Human Factors in Computing Systems (Hamburg, Germany) (CHI EA ’23). Association for Computing Machinery, New York, NY, USA, Article 413, 26 pages. doi:10.1145/3544549.3582749

    arXiv 2023

  4. [3]

    Thomas H Davenport and Nitin Mittal. 2023. All-in on AI: How Smart Companies Win Big with Artificial Intelligence . Harvard Business Review Press, Boston, MA

    work page 2023

  5. [4]

    Elizabeth Gibney. 2025. What are the best AI tools for research? Nature’s guide. Nature (17 Feb. 2025). doi:10.1038/d41586-025-00437-0

    work page doi:10.1038/d41586-025-00437-0 2025

  6. [5]

    Yen-Chia Hsu, Ting-Hao ‘Kenneth’ Huang, Himanshu Verma, Andrea Mauri, Illah Nourbakhsh, and Alessandro Bozzon. 2022. Empowering local communities using artificial intelligence. Patterns 3, 3 (March 2022), 100449. doi:10.1016/j.patter.2022.100449

    work page arXiv 2022

  7. [6]

    Jason T. Jacques. 2020. CHI 2020: Right Here, Right Now? A bottom-up approach to estimating the carbon emissions from more than twenty years of CHI conference travel.. In Extended Abstracts of the 2020 CHI Conference on Human Factors in Computing Systems (Honolulu, HI, USA) (CHI EA ’20). Association for Computing Machinery, New York, NY, USA, 1–13. doi:10...

    work page arXiv 2020

  8. [7]

    Annika Kaltenhauser, Gian-Luca Savino, Nick von Felten, and Johannes Schöning. 2025. CHI’s Greatest Hits: Analyzing the 100 Most-Cited Papers in 43 Years of Research at ACM CHI. Interactions 32, 1 (Jan. 2025), 28–33. doi:10.1145/3704804

    work page doi:10.1145/3704804 2025

Show all 27 references
  1. [8]

    Hao-Ping (Hank) Lee, Advait Sarkar, Lev Tankelevitch, Ian Drosos, Sean Rintel, Richard Banks, and Nicholas Wilson. 2025. The Impact of Generative Al on Critical Thinking: Self-Reported Reductions in Cognitive Effort and Confidence Effects From a Survey of Knowledge Workers. In...

    2025

  2. [9]

    Weixin Liang, Yaohui Zhang, Zhengxuan Wu, Haley Lepp, Wenlong Ji, Xuandong Zhao, Hancheng Cao, Sheng Liu, Siyu He, Zhi Huang, Diyi Yang, Christopher Potts, Christopher D Manning, and James Y. Zou. 2024. Mapping the Increasing Use of LLMs in Scientific Papers. doi:10.48550/arXi...

    2024 doi arXiv

  3. [10]

    Alessandro Mantelero. 2022. Beyond Data: Human Rights, Ethical and Social Impact Assessment in AI . T.M.C. Asser Press. doi:10.1007/978-94-6265- 531-7

    2022 doi

  4. [11]

    Abraham Mhaidli and Kat Roemmich. 2024. Overworking in HCI: A Reflection on Why We Are Burned Out, Stressed, and Out of Control; and What We Can Do About It. In Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (Honolulu, HI, USA) (CHI EA ’24). Ass...

    2024

  5. [12]

    Neil Millar, Brian Budgell, and Françoise Salager-Meyer. 2020. Hype in reports of clinical research: The authors’ perspectives. English for Specific Purposes 60 (Oct. 2020), 53–64. doi:10.1016/j.esp.2020.07.001

    2020 doi

  6. [13]

    Melanie Mitchell. 2019. Artificial Intelligence: A Guide for Thinking Humans . Pelican, London, England

    2019

  7. [14]

    Sriraam Natarajan, Saurabh Mathur, Sahil Sidheekh, Wolfgang Stammer, and Kristian Kersting. 2024. Human-in-the-loop or AI-in-the-loop? Automate or Collaborate? arXiv:2412.14232 [cs.HC] https://arxiv.org/abs/2412.14232 [Preprint]

    2024 arXiv

  8. [15]

    Rock Yuren Pang, Hope Schroeder, Kynnedy Simone Smith, Solon Barocas, Ziang Xiao, Emily Tseng, and Danielle Bragg. 2025. Understanding the LLM-ification of CHI: Unpacking the Impact of LLMs at CHI through a Systematic Literature Review. In CHI Conference on Human Factors in Co...

    2025 doi arXiv

  9. [16]

    Williams, Karin Hansson, and Ivana Feldfeber

    Gisela Reyes-Cruz, Velvet Spors, Michael Mueller, Marianela Ciolfi Felice, Shaowen Barzell, Rua M. Williams, Karin Hansson, and Ivana Feldfeber

  10. [18]

    Katja Rogers, Teresa Hirzle, Sukran Karaosmanoglu, Paula Toledo Palomino, Ekaterina Durmanova, Seiji Isotani, and Lennart E. Nacke. 2024. An Umbrella Review of Reporting Quality in CHI Systematic Reviews: Guiding Questions and Best Practices for HCI. ACM Trans. Comput.-Hum. In...

    2024 doi

  11. [19]

    Albrecht Schmidt, Passant Elagroudy, Fiona Draxler, Frauke Kreuter, and Robin Welsch. 2024. Simulating the Human in HCD with ChatGPT: Redesigning Interaction Design with AI. Interactions 31, 1 (Jan. 2024), 24–31. doi:10.1145/3637436

    2024 doi

  12. [20]

    Robert Skidelsky. 2017. Death to machines?: By Robert Skidelsky. https://www.project-syndicate.org/commentary/robert-skidelsky-revisits-the- luddites--claim-that-automation-depresses-real-wages

    2017

  13. [21]

    Nenad Tomašev, Julien Cornebise, Frank Hutter, Shakir Mohamed, Angela Picciariello, Bec Connelly, Danielle C. M. Belgrave, Daphne Ezer, Fanny Cachat van der Haert, Frank Mugisha, Gerald Abila, Hiromi Arai, Hisham Almiraat, Julia Proskurnia, Kyle Snyder, Mihoko Otake-Matsuura, ...

  14. [22]

    Dominik Vrabič Dežman. 2024. Promising the future, encoding the past: AI hype and public media imagery. AI and Ethics 4, 3 (April 2024), 743–756. doi:10.1007/s43681-024-00474-x Manuscript submitted to ACM Emphasizing Deliberation and Critical Thinking in an AI Hype World 5

    2024 doi

  15. [23]

    Jess Whittlestone and Samuel Clarke. 2022. AI challenges for society and ethics. In The Oxford Handbook of AI Governance . Oxford University Press. https://arxiv.org/pdf/2206.11068

    2022 arXiv

  16. [24]

    Mikael Wiberg and Erik Stolterman. 2014. What makes a prototype novel? a knowledge contribution concern for interaction design research. In Proceedings of the 8th Nordic Conference on Human-Computer Interaction: Fun, Fast, Foundational (Helsinki, Finland) (NordiCHI ’14). Assoc...

    2014

  17. [25]

    Carole-Jean Wu, Ramya Raghavendra, Udit Gupta, Bilge Acun, Newsha Ardalani, Kiwan Maeng, Gloria Chang, Fiona Aga, Jinshi Huang, Charles Bai, Michael Gschwind, Anurag Gupta, Myle Ott, Anastasia Melnikov, Salvatore Candido, David Brooks, Geeta Chauhan, Benjamin Lee, Hsien- Hsin ...

    2022

  18. [2020]

    Nature Communications 11, 1 (May 2020)

    AI for social good: unlocking the opportunity for positive impact. Nature Communications 11, 1 (May 2020). doi:10.1038/s41467-020-15871-z

    2020 doi

  19. [2025]

    In Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’24)

    Resisting Al Solutionism: Where Do We Go From Here?. In Extended Abstracts of the CHI Conference on Human Factors in Computing Systems (CHI EA ’24). doi:10.1145/3706599.3706732

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.