The DOI matches its Crossref record; 136 of the 136 references it cites were found in a registry or an index; 9 of the 117 sources its claims cite were checked for support.
Towards Human-AI Deliberation: Design and Evaluation of LLM-Empowered Deliberative AI for AI-Assisted Decision-Making
Abstract
Traditional AI-assisted decision-making systems often provide fixed recommendations that users must either accept or reject entirely, limiting meaningful interaction - especially in cases of disagreement. To address this, we introduce Human-AI Deliberation, an approach inspired by human deliberation theories that enables dimension-level opinion elicitation, iterative decision updates, and structured discussions between humans and AI. At the core of this approach is Deliberative AI, an assistant powered by large language models (LLMs) that facilitates flexible, conversational interactions and precise information exchange with domain-specific models. Through a mixed-methods user study, we found that Deliberative AI outperforms traditional explainable AI (XAI) systems by fostering appropriate human reliance and improving task performance. By analyzing participant perceptions, user experience, and open-ended feedback, we highlight key findings, discuss potential concerns, and explore the broader applicability of this approach for future AI-assisted decision-making systems.
Overview
CoChat Summary
TL;DR
The paper proposes Human-AI Deliberation, an LLM-empowered Deliberative AI that integrates domain-specific models with LLMs to support dimension-level opinion elicitation, iterative updates, and structured human-AI discussion in AI-assisted decision-making. In an exploratory graduate-admissions study, Deliberative AI improved decision accuracy and reduced over-reliance compared with traditional XAI, though it lowered trust and satisfaction.
Key findings
- Decision accuracy was significantly higher in Deliberative AI (M=0.598, SD=0.169) than in XAI (M=0.524, SD=0.16, p<0.05).
- Deliberative AI showed lower agreement fraction (M=0.57, SD=0.24 vs. M=0.68, SD=0.27, p<0.05) and lower switch fraction (M=0.23, SD=0.35 vs. M=0.51, SD=0.41, p<0.001) than XAI; over-reliance was lower (M=0.47, SD=0.31 vs. M=0.65, SD=0.33, p<0.001), while under-reliance did not differ significantly (M=0.32, SD=0.28 vs. M=0.29, SD=0.30).
- Deliberative AI participants reported lower trust than XAI participants (M=4.47, SD=1.68 vs. M=5.52, SD=1.27, p<0.01); no significant differences were found for helpfulness, understanding, mental demand, effort, or complexity, but satisfaction was significantly lower in Deliberative AI than in Human Alone. XAI participants reported higher confidence than Human Alone (M=6, SD=1.10 vs. M=5.59, SD=1.12, p<0.05) despite XAI final accuracy being lower than Human Alone.
- Qualitatively, 43/48 Deliberative AI participants felt deliberation helped them make more informed decisions; 21/48 identified AI limitations, 15/48 reflected on and corrected their own mistakes, and 18/48 gained new knowledge. For AI updates, 25/48 said updates improved user experience, but 11/48 felt the AI updated too frequently and 13/48 were uncertain why or how the AI was updating.
Methods
- Between-subjects mixed-methods study on Prolific: 174 recruited, 153 valid responses (Deliberative AI: 48, XAI: 51, Human Alone: 54). Participants had US graduate-admission experience and completed four main graduate-admission cases plus a tutorial.
- The task used a synthesized dataset of 100 applicant profiles; a domain-specific linear regression classifier achieved 80% accuracy on a 30% test set. Study cases were selected from the test set and intentionally chosen to be ambiguous/conflict-prone, with human and AI individual performance between 50% and 60%.
- Quantitative analysis used non-parametric tests (Mann-Whitney U; Kruskal-Wallis with Bonferroni correction); qualitative analysis used inductive thematic analysis of open-ended feedback and conversation logs. Pilot work included 30 participants and 226 deliberative statements; the Intention Analyzer reached 96% accuracy, and the Argument Evaluator showed Cohen’s κ=0.78 agreement with human scores.
Limitations
- Stated: the college admissions task does not fully capture real-world admissions; SOP and recommendation-letter strength are reduced to scale values; participants were not actual admissions committee members; most had master’s rather than PhD admissions experience, so findings may not generalize to PhD admissions.
- Stated: graduate admissions is subjective and lacks definitive ground truth; dataset labels came from a professional admissions committee, and accuracy plus over/under-reliance were used as objective metrics “to some extent.” The study used only four main tasks due to pilot fatigue, and the selected cases were intentionally ambiguous with low independent human/AI accuracy (50–60%), leaving less-ambiguous cases unaddressed.
- Evident/mentioned: the AI opinion-update mechanism was not disclosed to participants, and some doubted its transparency or reliability; the LLM was GPT-3.5 and the domain model was a linear regression classifier, limiting claims about other models or tasks; the authors describe the study as exploratory rather than an effectiveness test for graduate admissions.
Model-written summary. Check it against the paper before relying on it.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
3 Valid · 3 Not valid · 51 Not checked · 137 Own results and interpretations
At least one cited source doesn’t support this claim or points to a different paper, and none confirmed it — see the verdicts below.
In the paper
“However, unlike in [16], where cognitive effort significantly increased under-reliance (i.e., humans under high cognitive effort may blindly ignore AI's suggestions), our findings reveal no such adverse effect.”
What the sources say
- Disputedcites [16] [16] Zana Buçinca, Maja Barbara Malaya, and Krzysztof Z Gajos. 2021. To trust or to think: cognitive forcing functions can reduce overreliance on AI in AI-assisted decision-making. Proceedings of the ACM on Human-Computer Interaction 5, CSCW1 (2021), 1-21. - sourceExperimental — accuracy not yet measured
Our copy of Buçinca et al. (2021) reports the opposite of what the citing sentence attributes to it: in Section 4.1 (Objective measures), participants in the cognitive forcing conditions 'overrelied less' and 'there were no significant differences between categories for human errors' (both for overall performance and for carb source detection, where CFF participants 'overrelied significantly less'). The Discussion (Sec. 6) likewise states that cognitive forcing functions 'reduced overreliance on AI compared to the simple explainable AI approaches', with the trade-off being reduced perceived usability/trust/preference, not increased disregard of correct AI suggestions. Nowhere does our copy report that cognitive effort significantly increased under-reliance; the adverse effect the citing paper contrasts against is not this paper's finding. (LLM, conf=0.78)
Checked by a reasoning model — may be wrong.
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Disputed
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“Second, instead of full consensus or complete divergence, human and AI decision rationales often exhibit partial alignment [113, 128].”
What the sources say
- Supportedcites [113] [113] Venkatesh Sivaraman, Leigh A Bukowski, Joel Levin, Jeremy M Kahn, and Adam Perer. 2023. Ignore, trust, or negotiate: understanding clinician acceptance of AI-based treatment recommendations in health care. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems . 1-18. - source
Section 5.2.2 (Negotiate) states that 'the most common behavior pattern we observed was of participants selectively adopting aspects of the recommendation', arriving at 'an intermediate solution', and Fig. 6 contrasts this with the Ignore All / Accept All extremes — i.e. neither full consensus nor complete divergence. Section 5.2 (Patterns of Interaction) reinforces this quantitatively: actions 'did not perfectly align with the AI either' (only 6 of 36 discordant-with-attending decisions showed full concordance with the AI), and Section 6.1 describes 'negotiated partial forms of reliance'. The one qualification is that our copy frames the alignment in terms of clinicians' treatment decisions and their think-aloud reasoning in a single ICU sepsis task, not 'rationales' as such across human–AI decisions generally; the citing sentence generalizes slightly beyond that setting. (LLM, conf=0.82)
- Partly supportedcites [128] [128] Xinru Wang and Ming Yin. 2023. Watch Out for Updates: Understanding the Effects of Model Explanation Updates in AI-Assisted Decision Making. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems . 1-19. - sourceExperimental — accuracy not yet measured
Our copy treats human-AI alignment as a matter of degree rather than all-or-nothing: the Introduction (Sec. 1) writes that 'the more its explanations align with the human rationale, the more accurate users perceive the model to be', and Sec. 5.4 acknowledges that agreement fraction 'may reflect the natural agreement between people's independent decisions and the AI recommendations to some extent' — i.e. partial, not total, convergence. However, the paper never states that human and AI decision rationales 'often exhibit partial alignment rather than full consensus or complete divergence'; the rationale-alignment point it does make is explicitly credited to prior work ([63]), and the study's own similarity manipulations (Sec. 4.2.2, feature agreement scores 1.0/0.5/0.07) concern the old vs. updated AI explanations, not human vs. AI rationales. The 'decisions' rather than 'rationales' framing and the absent frequency claim ('often') are the parts not supported by this source. (LLM, conf=0.65)
plainstates an association- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Supported
The sources with a verdict support this claim; the rest weren’t checked yet, have no full text we could read, or their verdicts aren’t rated yet.
In the paper
“First, a battery of empirical studies found that people rarely trigger analytical thinking when directly presented with AI's suggestions [11, 15, 105].”
What the sources say
- Not checked yetcites [11] [11] Astrid Bertrand, Rafik Belloum, James R Eagan, and Winston Maxwell. 2022. How Cognitive Biases Affect XAI-assisted Decision-making: A Systematic Review. In Proceedings of the 2022 AAAI/ACM Conference on AI, Ethics, and Society . 78-91. - source
Checked before — its earlier verdict is held back until a new check runs.
- Supportedcites [15] [15] Zana Buçinca, Phoebe Lin, Krzysztof Z Gajos, and Elena L Glassman. 2020. Proxy tasks and subjective measures can be misleading in evaluating explainable AI systems. In Proceedings of the 25th international conference on intelligent user interfaces . 454-464. - source
The Discussion (p. 9) states outright that in actual decision-making situations people often choose not to engage in effortful analytical thinking, and that in the paper's own actual-task condition participants 'could complete the task even without engaging with the explanations' and 'chose to over-rely on the AI's recommendation, just to avoid cognitive effort'. The Introduction (p. 1) makes the same point — people favor heuristics over analytical thinking — and Section 2.1 (p. 2) reports that engagement with a system's recommendation required a cognitive forcing strategy (Green & Chen), i.e. it does not happen spontaneously when the AI's suggestion is simply shown. Together these cover the claim's substance, including its AI-suggestion specificity; the only hedge ('we suspect') attaches to the mechanism behind the over-reliance, not to the reluctance to think analytically itself. (LLM, conf=0.85)
- Supportedcites [105] [105] Charvi Rastogi, Yunfeng Zhang, Dennis Wei, Kush R Varshney, Amit Dhurandhar, and Richard Tomsett. 2022. Deciding fast and slow: The role of cognitive biases in ai-assisted decision-making. Proceedings of the ACM on Human-Computer Interaction 6, CSCW1 (2022), 1-22. - source
Section 4 (Anchoring Bias) states that in AI-assisted decision-making 'the human decision-maker is irrationally anchored to the AI-generated decision' and that 'after being anchored, humans tend to adjust insufficiently because adjustments are effortful and tend to stop once a plausible estimate is reached'; the Introduction adds that anchoring 'manifests through blind reliance on the anchor... which limits the exploration of alternative hypotheses'. The paper's own Experiment 1 (Sec. 4.1) reports that 'the likelihood of disagreement for unmodified trials is low (close to 0.1)', i.e. participants almost always accepted the presented AI prediction, and Sec. 5.3 reports agreement percentages of ~84-88% in the collaborative conditions. Related Work (Sec. 2) likewise relays that users 'display over-reliance on AI due to blind trust in automation' and 'do not calibrate their reliance on AI based on its accuracy'. Together these carry the claim's content — people seldom engage the effortful, analytical adjustment once an AI suggestion is in front of them — so the paraphrase is supported; the only wording difference is that the paper speaks of anchoring/insufficient adjustment and heuristic thinking rather than 'triggering analytical thinking'. (LLM, conf=0.82)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Not checked yet
The cited sources support only part of this claim — read each source’s verdict below.
In the paper
“However, existing research has found that when people are provided with a recommendation and passively look at it, they often lack analytical thinking, leading to over-reliance or under-reliance on AI systems [15, 42, 64, 79].”
What the sources say
- Partly supportedcites [15] [15] Zana Buçinca, Phoebe Lin, Krzysztof Z Gajos, and Elena L Glassman. 2020. Proxy tasks and subjective measures can be misleading in evaluating explainable AI systems. In Proceedings of the 25th international conference on intelligent user interfaces . 454-464. - sourceExperimental — accuracy not yet measured
Our copy supports most of the paraphrase: the Discussion (p. 9) states that 'in actual decision-making situations, people often choose not to engage in effortful analytical thinking, even in high-stakes situations like medical diagnosis', and the Introduction (p. 1) that people 'favor less demanding forms of cognition, i.e., heuristics over analytical thinking' when their cognitive focus is elsewhere rather than on the AI (the passive-viewing situation the proxy/actual-task contrast describes). The over-reliance half is also there: the Discussion (p. 9) infers participants 'chose to over-rely on the AI's recommendation, just to avoid cognitive effort of examining those explanations'. However, the paired 'under-reliance' is not attributed to a lack of analytical thinking in this paper — the front matter (p. 1) presents it as the opposite end of the trust spectrum ('they might exhibit self-reliance and ignore the system's correct recommendations'), a trust-calibration problem rather than a consequence of disengagement, and the paper's own under-reliance-type result (participants ignoring incorrect AI recommendations, Discussion p. 9) is tied to explanation type, not to effort avoidance. Also note the paper hedges its causal link ('we suspect'), while the citing sentence reports it as a finding of existing research, and credits the analytical-thinking evidence to other works ([24, 25, 31]). (LLM, conf=0.72)
- Partly supportedcites [42] [42] Krzysztof Z Gajos and Lena Mamykina. 2022. Do People Engage Cognitively with AI? Impact of AI Assistance on Incidental Learning. In 27th International Conference on Intelligent User Interfaces . 794-806. - sourceExperimental — accuracy not yet measured
Gajos & Mamykina (2022) supports the over-reliance half of the claim: the Introduction (p. 1) states that previous studies showed a human tendency to over-rely on AI-generated recommendations, 'which suggests that people may be processing information provided by AI superficially rather than engaging with it deeply and critically,' and the Related Work (p. 2) reports 'converging evidence that people overrely on the AI-generated recommendations'; the Discussion (p. 11) adds 'compelling evidence ... that people do not carefully process the AI-generated information when given a decision recommendation and an explanation.' However, no passage in our copy attributes under-reliance to this line of work — the paper discusses only over-reliance (and interventions that reduce it). Also, the source frames the superficial-processing account as a suggestion/hypothesis rather than an established finding, whereas the citing sentence states it as what research 'has found.' (LLM, conf=0.78)
- Not checked yetcites [64] [64] Harmanpreet Kaur, Harsha Nori, Samuel Jenkins, Rich Caruana, Hanna Wallach, and Jennifer Wortman Vaughan. 2020. Interpreting interpretability: understanding data scientists' use of interpretability tools for machine learning. In Proceedings of the 2020 CHI conference on human factors in computing systems . 1-14. - source
- Partly supportedcites [79] [79] QVera Liao and Kush R Varshney. 2021. Human-Centered Explainable AI (XAI): From Algorithms to User Experiences. arXiv preprint arXiv:2110.10790 (2021). - sourceExperimental — accuracy not yet measured
Section 4 (p. 10) of our copy supports the cognitive core of the claim: via dual-process theories, 'people are more likely to engage in System 1 thinking' rather than attending to every bit of explanations, and people resort to System 1 when they 'lack either the ability or motivation ... to perform analytical thinking'. The reliance consequences are present but hedged and framed as possibilities/risks rather than established findings — 'It is possible that people superficially associate the ability to provide explanations directly with competence, and therefore form unwarranted trust and confidence' (p. 10) and 'the risk of mistrust and misuse of AI' (p. 10), with 'appropriate reliance' discussed only as a desideratum (p. 8). The citing sentence therefore states as a settled finding what this source presents as an emerging awareness, and the source's account concerns processing of AI explanations/recommendations in XAI contexts specifically. (LLM, conf=0.75)
plainstates a cause- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: passes — The paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Partly supported
The cited sources that could be checked couldn’t confirm or dispute this claim.
In the paper
“Despite this, engaging users with strong subjective perspectives remains challenging [80].”
What the sources say
- Could not verifycites [80] [80] Hanmeng Liu, Ruoxi Ning, Zhiyang Teng, Jian Liu, Qiji Zhou, and Yue Zhang. 2023. Evaluating the logical reasoning ability of chatgpt and gpt-4. arXiv preprint arXiv:2304.03439 (2023). - sourceExperimental — accuracy not yet measured
Our copy of Liu et al. (2023) is confined to logical-reasoning evaluation on NLI and multi-choice reading-comprehension benchmarks (Sec. 2 Evaluation Settings; Secs. 3.1–3.2 results; Sec. 4.1 error analysis, 4.2 in-context learning, 4.3 chain-of-thought), and its stated limitations concern out-of-distribution datasets and instruction-following, not user engagement or subjective perspectives. Searched for 'subjective perspective / stance / opinion / user engagement / interactivity' and for conversational/empathetic limitations: no passage in our copy addresses engaging users holding strong subjective viewpoints. Nothing contrary is said either, so the claim is simply outside this source's scope. (LLM, conf=0.72)
plain- Asserted: passes — The authors put it forward as true.
- Checkable: passes — It can be checked against evidence.
- Load-bearing: unclear — It is unclear whether the paper’s conclusions depend on it.
- Based: passes — The paper gives a reason why it holds.
- Could not verify
73 more claims
Create a free account to see all 194 claims and to verify the 103 references still open.
- It may state two separate claims as one. 68
- Probably not a substantial claim. 75
Citations
Cites
Three Approaches to Qualitative Content Analysis
Hsieh, H.-F., & Shannon, S. E. (2005) · Qualitative Health Research · doi:10.1177/1049732305276687 · Cited by 44,601 · cited 2×
“For qualitative analysis, two authors independently coded participants’ open-ended feedback and conversation logs using an inductive thematic analysis approach [61].”
Statistical power analyses using G∗Power 3.1: tests for correlation and regression analyses
Faul, F., Erdfelder, E., Buchner, A., & Lang, A.-G. (2009) · Behavior Research Methods · doi:10.3758/brm.41.4.1149 · Cited by 37,615 · cited 1×
“Helpfulness “I think the AI model’s assistance is helpful/useful for me to make good decisions.”
Decision making with the analytic hierarchy process
Saaty, T. L. (2008) · International Journal of Services Sciences · doi:10.1504/ijssci.2008.017590 · Cited by 10,301 · cited 1×
“Finally, based on theories and practices in human de-liberation, we introduce the design considerations and structure of Deliberative AI , the core AI component of our framework.”
Trust in Automation: Designing for Appropriate Reliance
Lee, J. D., & See, K. A. (2004) · Human Factors: The Journal of the Human Factors and Ergonomics Society · doi:10.1518/hfes.46.1.50_30392 · Cited by 6,605 · cited 1×
“Both over-reliance, where individuals trust AI excessively [78, 104], and under-trust, where individuals fail to utilize AI when necessary [78], can lead to adverse outcomes.”
Explanation in artificial intelligence: Insights from the social sciences
Miller, T. (2019) · Artificial Intelligence · doi:10.1016/j.artint.2018.07.007 · Cited by 5,302
Survey of Hallucination in Natural Language Generation
Ji, Z., Lee, N., Frieske, R., Yu, T., Su, D., Xu, Y., Ishii, E., Bang, Y. J., Madotto, A., & Fung, P. (2023) · ACM Computing Surveys · doi:10.1145/3571730 · Cited by 5,025
Humans and automation: Use, misuse, disuse, abuse
Parasuraman, R., & Riley, V. (1997) · Human Factors: The Journal of the Human Factors and Ergonomics Society · doi:10.1518/001872097778543886 · Cited by 4,554 · cited 2×
“Both over-reliance, where individuals trust AI excessively [78, 104], and under-trust, where individuals fail to utilize AI when necessary [78], can lead to adverse outcomes.”
NASA-Task Load Index (NASA-TLX); 20 Years Later
Hart, S. G. (2006) · Proceedings of the Human Factors and Ergonomics Society Annual Meeting · doi:10.1177/154193120605000909 · Cited by 4,297 · cited 1×
“Meanwhile, as deliberation requires extra efforts, we evaluated participants’ self-reported Mental Demand [17, 46, 56, 71], Effort [56], Complexity [17], and Satisfaction [17, 46] via 7-point Likert scale.”
Models of Bounded Rationality: Empirically Grounded Economic Reason
Simon, H. A. (1997) · The MIT Press eBooks · doi:10.7551/mitpress/4711.001.0001 · Cited by 4,205
Retrieval-Augmented Generation for Knowledge-Intensive NLP Tasks
Lewis, P., Perez, E., Piktus, A., Petroni, F., Karpukhin, V., Goyal, N., Küttler, H., Lewis, M., Yih, W.-T., Rocktäschel, T., et al. (2020) · arXiv · doi:10.48550/arxiv.2005.11401 · Cited by 3,101 · cited 1×
“” (P33)”
Research works printed but not linked: 8
- André Bächtiger, Susumu Shikano, Seraina Pedrini, and Mirjam Ryser. 2009. Measuring deliberation 2.0: standards, discourse types, and sequenzialization. In ECPR General Conference. Potsdam, 5–12.
- Daniel Kahneman. 2011. Thinking, fast and slow. Macmillan.
- Scott M Lundberg and Su-In Lee. 2017. A unified approach to interpreting model predictions. Advances in neural information processing systems 30 (2017).
- Fred C Lunenburg. 2010. The decision making process.. In National Forum of Educational Administration & Supervision Journal , Vol. 27.
- Shuai Ma Mingfei Sun and Xiaojuan Ma. 2022. Modeling Adaptive Expression of Robot Learning Engagement and Exploring its Effects on Human Teachers. ACM Transactions on Computer-Human Interaction (2022).
- Philip E Tetlock. 2017. Expert political judgment. In Expert Political Judgment. Princeton University Press.
- Jason Wei Xuezhi Wang Dale Schuurmans Maarten Bosma Fei Xia Ed Chi Quoc V Le Denny Zhou et al. 2022. Chain-of-thought prompting elicits reasoning in large language models. Advances in neural information processing systems 35 (2022) 24824–24837.
- Jieqiong Zhao Yixuan Wang Michelle V Mancenido Erin K Chiou and Ross Maciejewski. 2023. Evaluating the impact of uncertainty visualization on model reliance. IEEE Transactions on Visualization and Computer Graphics (2023).
Cited by
Xiao, X., Chen, P., Qi, B., Zhao, H., Liang, J., Tong, J., & Wang, H. (2025) · Reliability Engineering & System Safety · doi:10.1016/j.ress.2025.111585 · Cited by 20
AI-overdependence and human cognitive decline: Hazards, evidence, and mitigation strategies
Noorbehbahani, F., & Oyibo, K. (2026) · Computers in Human Behavior Reports · doi:10.1016/j.chbr.2026.101102 · Cited by 12
LLMs in the SOC: An Empirical Study of Human-AI Collaboration in Security Operations Centres
Singh, R., Tariq, S., Jalalvand, F., Chhetri, M. B., Nepal, ., Paris, C., & Lochner, M. (2026) · IEEE Symposium on Security and Privacy (S&P) · doi:10.1109/sp63933.2026.00111 · Cited by 58
“Meanwhile, iterative reasoning and hypothesis testing interactions resemble the deliberative partner model [34], ExtendAI role [33] and align with Cooperative Intelligence [30].”
Raees, M., Khan, V.-J., Lykourentzou, I., & Papangelis, K. (2026) · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems · doi:10.1145/3772318.3791467 · Cited by 5
“Partial explanations align with a larger body of research on deliberation that encourages users to reflect on their decisions [40, 96].”
InfluentialDong, A., Li, P., Chen, Y., Gibson, S., Zhao, L., & He, M. (2026) · Sustainability · doi:10.3390/su18115313 · Cited by 8
Metadata
- Authors
- Ma, S., Chen, Q., Wang, X., Zheng, C., Peng, Z., Yin, M., & Ma, X.
- Affiliations
- Hong Kong University of Science and Technology (HK)
- Purdue University West Lafayette (US)
- Sun Yat-sen University (CN)
- Venue
- Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems
- Publisher
- ACM
- Published
- 2025-04-25
- Pages
- 1-23
- OpenAlex
- W4409736269
- Semantic Scholar
- 9eaf282a505a1d61318d4d90d7044da360ffb72f
- Cited by
- 48
- References
- 119
- Citations by year
- 2026: 42 · 2025: 6
- FWCI
- 45.52
- Type
- conference paper
- Language
- en
- Open access
- closed
- Locations
- CHI Conference on Human Factors in Computing Systems (publishedVersion)
- Rare & Special e-Zone (The Hong Kong University of Science and Technology) (submittedVersion)
- Funders
- Research Grants Council of the Hong Kong Special Administrative Region under General Research Fund (GRF) (16207923)
- Topics
- Physical Sciences › Computer Science › Artificial Intelligence › Explainable Artificial Intelligence (XAI)
- Physical Sciences › Computer Science › Artificial Intelligence › Ethics and Social Impacts of AI
- Social Sciences › Human Factors and Ergonomics › Human-Automation Interaction and Safety
- Keywords
- AI-assisted decision-making, large language models, human-AI interaction, explainable AI, conversational AI assistant, user experience
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
The claim verdicts this report counts are experimental: their accuracy is not yet measured.
A language model drew the claims from the paper on 2026-11-14, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).
A second model labelled each claim on 2026-11-14 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).
Cited sources are checked with a language model’s help (qwen3.8-omni-flash).
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
How to cite
Cite CoChat for the checks on this page, and the paper for its own findings.
- This page
- CoChat. (2026). Citation check: Towards Human-AI Deliberation: Design and Evaluation of LLM-Empowered Deliberative AI for AI-Assisted Decision-Making. https://app.cochat.ai/papers/10.1145/3706598.3713423
- The paper
- Ma, S., Chen, Q., Wang, X., Zheng, C., Peng, Z., Yin, M., & Ma, X. (2025). Towards Human-AI Deliberation: Design and Evaluation of LLM-Empowered Deliberative AI for AI-Assisted Decision-Making. Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3706598.3713423
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.