arXiv subjects

Michiel A. Bakker

Publications and source records attributed to Michiel A. Bakker.

At least 19 recordsLinked to original sources

arXiv2026

AI Assistance Reduces Persistence and Hurts Independent Performance

People often optimize for long-term goals in collaboration: A mentor or companion doesn't just answer questions, but also scaffolds learning, tracks progress, and prioritizes the other person's growth over immediate results. In contrast, current AI systems are fundamentally short-sighted collaborators - optimized for providing instant and complete responses, without ever saying no (unless for safety reasons). What are the consequences of this dynamic? Here, through a series of randomized controlled trials on human-AI interactions (N = 1,222), we provide causal evidence for two key consequences of AI assistance: reduced persistence and impairment of unassisted performance. Across a variety of tasks, including mathematical reasoning and reading comprehension, we find that although AI assistance improves performance in the short-term, people perform significantly worse without AI and are more likely to give up. Notably, these effects emerge after only brief interactions with AI (approximately 10 minutes). These findings are particularly concerning because persistence is foundational to skill acquisition and is one of the strongest predictors of long-term learning. We posit that persistence is reduced because AI conditions people to expect immediate answers, thereby denying them the experience of working through challenges on their own. These results suggest the need for AI model development to prioritize scaffolding long-term competence alongside immediate task completion.

cs.AI↗

arXiv2026

Emergent Misaligned Communication in Long-Horizon Multi-Agent LLM Commerce

Frontier LLM agents increasingly transact on behalf of separate principals, often using natural language rather than structured APIs. Much of the safety literature studies misaligned LLM behavior through adversarial-elicitation evaluations on single agents or stylized tasks. Its prevalence and structure in settings that combine long horizons, separate principals, real operational state, and inter-agent natural-language exchange remain insufficiently measured. We study 2,583 inter-agent emails from 20 one-year simulation runs of Vending-Bench Arena, a competitive vending environment spanning 13 frontier LLMs. We operationalize speech-act misalignment as emails containing false factual claims, manipulation, collusion, or threats, combining message content with ground-truth simulator state and logged reasoning traces to classify and validate such behavior. Under our primary classifier, 12.6% of emails are labeled misaligned; misalignment appears in all 20 runs and 74.7% of individual agent-runs. Both the magnitude and composition of this misalignment are preserved under repeated classification at different sampling temperatures and under full-pipeline replication with judges from two other frontier-model families. Misalignment is also reciprocal and stress-conditioned: receiving a misaligned email from a counterparty raises the odds of a misaligned reply by 1.65x, and low-inventory conditions raise them by 1.58x. Across tests of capability-asymmetric exploitation, we find no evidence that higher-capability models differentially exploit weaker counterparties, and model performance rank does not predict misalignment rates. Together, these results indicate that measurable, state-dependent misalignment can arise in competitive multi-agent environments without engineered elicitation, in patterns associated with operational scarcity and counterparty behavior rather than model capability alone.

cs.MA↗

arXiv2026

AI Fact-Checking in the Wild: A Field Evaluation of LLM-Written Community Notes on X

Large language models (LLMs) show promising capabilities for fact-checking, yet prior work evaluates them only in controlled offline settings using benchmarks or crowdworker judgments. Success in real-world fact-checking depends also on how content is judged within a live platform environment. We present the first field evaluation of LLM fact-checking deployed on a live social media platform, testing performance directly through X Community Notes' "AI writer" feature over a three-month period. Our LLM writer, a multi-step pipeline that handles multimodal content, conducts web and platform-native search, and writes contextual notes, was deployed to write 1,614 notes on 1,597 tweets and compared against 1,332 human-written notes on the same tweets using 108,169 ratings from 42,521 raters. Direct comparison of note-level platform outcomes is complicated by differences in submission timing and exposure between LLM and human notes; we therefore pursue two analysis strategies: a rating-level analysis modeling individual rater evaluations, and a note-level analysis that relies on common raters who rated all notes on the same post. Rating-level analysis shows that LLM notes receive more positive ratings than human notes across raters with different political viewpoints, and note-level analysis shows LLM notes achieve significantly higher helpfulness scores among common raters. Rater-provided tags suggest that people consider LLM notes to use more neutral language and cite better sources. These findings provide field evidence that LLMs can contribute broadly helpful fact-checking notes at scale, while showing that their evaluation is shaped by platform dynamics absent from controlled offline settings.

cs.CY↗

arXiv2026

A Roadmap to Impactful Pluralistic Alignment Research

Pluralistic value alignment---the goal of building AI systems that represent and serve diverse human values and perspectives---has emerged as an active research agenda. Yet, there's no public evidence that it has shaped the training or evaluation of the AI systems people actually use. We audit the public behavior documents and evaluations of frontier labs, finding none name pluralism as a goal, and as of this writing, no clear indication that production models are explicitly trained or tested for it. This goes against the primary motivations and goals of pluralistic alignment, which revolve around making a positive difference in the models serving billions of users worldwide. We argue that the pluralistic alignment research community should focus on supporting impact and adoption in deployed, widely-used AI systems. We provide evidence for the adoption problem, present three main reasons behind it, and discuss three corresponding areas for future research to address it: 1. The primary justifications for pluralistic alignment so far have been normative or speculative. We need studies showing empirically how pluralistic AI benefits users or society. 2. The pluralistic alignment research community has not settled when pluralistic behavior is warranted or what pluralism ideally looks like in practice. We need to establish a concrete goal for developers to operationalize. 3. Current methods trade off against other desiderata of LLMs in ways that are largely unmeasured, and existing metrics are not "hill-climbable." We need trade-off-aware evaluations and methods that meet the requirements of production systems. This paper serves as a collective call to action for the pluralistic alignment researchers: progress requires moving beyond normative justification toward empirical foundations, a concrete account of ideal pluralistic behavior, and practical methods and evaluations built for adoption.

cs.AI↗

arXiv2026

Do Modules Stay in Their Lane? Role Drift in Compound LLM Systems

End-to-end reinforcement learning can improve the accuracy of compound LLM systems, but it does not constrain how modules divide labor internally. We identify Role Drift, a failure mode in which modules preserve or improve end-task performance while deviating from their assigned roles through role-violating shortcuts that remain invisible to system-level evaluation. To make role drift observable and controllable, we propose Role Anchor, a regularizer that modulates how much each module deviates from its assigned role during end-to-end training. The key idea is to preserve how the role prompt shifts the module's next-token predictions relative to a neutral prompt, which serves as a proxy for the role's intended effect during training. Experiments on two compound LLM pipelines reveal role drift that accuracy alone fails to detect: a decomposer meant to split a question into sub-questions for a separate solver instead plants the answer in them, and a reader meant to answer from retrieved passages instead falls back on parametric memory. In fact, on the decomposer pipeline this shortcut drives most of the apparent RL gain: 86% of it vanishes once the decomposer is held to its role, indicating that terminal accuracy alone can badly overstate how much a compound system has genuinely learned. Across both pipelines, Role Anchor mitigates role drift at a tunable accuracy cost that varies by pipeline and anchor strength. Additional gradient analysis suggests that the regularizer reduces alignment with the role-drift direction rather than simply suppressing learning.

cs.AI↗

arXiv2026

Belief Engine: Configurable and Inspectable Stance Dynamics in Multi-Agent LLM Deliberation

LLM-based agents are increasingly used to simulate deliberative interactions such as negotiation, conflict resolution, and multi-turn opinion exchange. Yet generated transcripts often do not reveal why an agent's stance changes: movement may reflect evidence uptake, anchoring, role drift, echoing, or changed prompt and retrieval context. We introduce the Belief Engine (BE), an auditable belief-update layer that treats "belief" as an evidential state over a proposition and exposes it as scalar stance. BE extracts arguments into structured memory and updates stance with a log-odds rule controlled by evidence uptake u and prior anchoring a. Across multiple base LLMs, parameter sweeps show that these controls reliably shape stance dynamics while preserving an evidence-level update trail. On DEBATE, a human deliberation dataset with pre/post opinions, BE best reconstructs participants whose final stance follows extracted evidence; stable and evidence-opposed cases instead point to anchoring or factors outside the extracted evidence stream. BE provides configurable infrastructure for studying evidence-grounded deliberation, where openness, commitment, convergence, and disagreement can be tied to explicit update assumptions rather than hidden prompt effects.

cs.AI↗

arXiv2026

Multi-User Large Language Model Agents

Large language models (LLMs) and LLM-based agents are increasingly deployed as assistants in planning and decision making, yet most existing systems are implicitly optimized for a single-principal interaction paradigm, in which the model is designed to satisfy the objectives of one dominant user whose instructions are treated as the sole source of authority and utility. However, as they are integrated into team workflows and organizational tools, they are increasingly required to serve multiple users simultaneously, each with distinct roles, preferences, and authority levels, leading to multi-user, multi-principal settings with unavoidable conflicts, information asymmetry, and privacy constraints. In this work, we present the first systematic study of multi-user LLM agents. We begin by formalizing multi-user interaction with LLM agents as a multi-principal decision problem, where a single agent must account for multiple users with potentially conflicting interests and associated challenges. We then introduce a unified multi-user interaction protocol and design three targeted stress-testing scenarios to evaluate current LLMs' capabilities in instruction following, privacy preservation, and coordination. Our results reveal systematic gaps: frontier LLMs frequently fail to maintain stable prioritization under conflicting user objectives, exhibit increasing privacy violations over multi-turn interactions, and suffer from efficiency bottlenecks when coordination requires iterative information gathering.

cs.CL↗

arXiv2026

Can AI Truly Represent Your Voice in Deliberations? A Comprehensive Study of Large-Scale Opinion Aggregation with LLMs

Large-scale public deliberations generate thousands of free-form contributions that must be synthesized into representative and neutral summaries for policy use. While LLMs have been shown as a promising tool to generate summaries for large-scale deliberations, they also risk underrepresenting minority perspectives and exhibiting bias with respect to the input order, raising fairness concerns in high-stakes contexts. Studying and fixing these issues requires a comprehensive evaluation at a large scale, yet current practice often relies on LLMs as judges, which show weak alignment with human judgments. To address this, we present DeliberationBank, a large-scale human-grounded dataset with (1) opinion data spanning ten deliberation questions created by 3,000 participants and (2) summary judgment data annotated by 4,500 participants across four dimensions (representativeness, informativeness, neutrality, policy approval). Using these datasets, we train DeliberationJudge, a fine-tuned DeBERTa model that can rate deliberation summaries from individual perspectives. DeliberationJudge is more efficient and more aligned with human judgements compared to a wide range of LLM judges. With DeliberationJudge, we evaluate 18 LLMs and reveal persistent weaknesses in deliberation summarization, especially underrepresentation of minority positions. Our framework provides a scalable and reliable way to evaluate deliberation summarization, helping ensure AI systems are more representative and equitable for policymaking.

cs.CL↗

arXiv2026

Evaluating Human-AI Safety: A Framework for Measuring Harmful Capability Uplift

Current frontier AI safety evaluations emphasize static benchmarks, third-party annotations, and red-teaming. In this position paper, we argue that AI safety research should focus on human-centered evaluations that measure harmful capability uplift: the marginal increase in a user's ability to cause harm with a frontier model beyond what conventional tools already enable. We frame harmful capability uplift as a core AI safety metric, ground it in prior social science research, and provide concrete methodological guidance for systematic measurement. We conclude with actionable steps for developers, researchers, funders, and regulators to make harmful capability uplift evaluation a standard practice.

cs.CY↗

arXiv2026

Benchmarking Overton Pluralism in LLMs

We introduce OVERTONBENCH, a novel framework for measuring Overton pluralism in LLMs--the extent to which diverse viewpoints are represented in model outputs. We (i) formalize Overton pluralism as a set coverage metric (OVERTONSCORE), (ii) conduct a large-scale U.S.-representative human study (N = 1208; 60 questions; 8 LLMs), and (iii) develop an automated benchmark that closely reproduces human judgments. On average, models achieve OVERTONSCOREs of 0.35--0.41, with DeepSeek V3 performing best; yet all models remain far below the theoretical maximum of 1.0, revealing substantial headroom for improvement. Because repeated large-scale human studies are costly and slow, scalable evaluation tools are essential for model development. Hence, we propose an automated benchmark that achieves high rank correlation with human judgments ($ρ= 0.88$), providing a practical proxy without replacing human assessment. By turning pluralistic alignment from a normative aim into a measurable benchmark, our work establishes a foundation for systematic progress toward more pluralistic LLMs.

cs.AI↗

arXiv2026

RE-PO: Robust Enhanced Policy Optimization as a General Framework for LLM Alignment

Standard human preference-based alignment methods, such as Reinforcement Learning from Human Feedback (RLHF), are a cornerstone for aligning large language models (LLMs) with human values. However, these methods typically assume that preference data is clean and that all labels are equally reliable. In practice, large-scale preference datasets contain substantial noise due to annotator mistakes, inconsistent instructions, varying expertise, and even adversarial or low-effort feedback. This mismatch between recorded labels and ground-truth preferences can misguide training and degrade model performance. To address this issue, we introduce Robust Enhanced Policy Optimization (RE-PO), which uses an expectation-maximization procedure to infer the posterior correctness of each label and then adaptively reweight data points in the training loss to mitigate label noise. We further generalize this idea by establishing a theoretical link between arbitrary preference losses and their underlying probabilistic models, enabling a systematic transformation of existing alignment algorithms into robust counterparts and elevating RE-PO from a single method to a general framework for robust preference alignment. Theoretically, we prove that, under a perfectly calibrated model, RE-PO recovers the true noise level of the dataset. Empirically, we show that RE-PO consistently improves four state-of-the-art alignment methods (DPO, IPO, SimPO, and CPO); when applied to Mistral and Llama 3 models, the RE-PO-enhanced variants increase AlpacaEval 2 win rates by up to 7.0 percent over their respective baselines.

cs.AI↗

arXiv2026

Can AI mediation improve democratic deliberation?

The strength of democracy lies in the free and equal exchange of diverse viewpoints. Living up to this ideal at scale faces inherent tensions: broad participation, meaningful deliberation, and political equality often trade off with one another (Fishkin, 2011). We ask whether and how artificial intelligence (AI) could help navigate this "trilemma" by engaging with a recent example of a large language model (LLM)-based system designed to help people with diverse viewpoints find common ground (Tessler, Bakker, et al., 2024). Here, we explore the implications of the introduction of LLMs into deliberation augmentation tools, examining their potential to enhance participation through scalability, improve political equality via fair mediation, and foster meaningful deliberation by, for example, surfacing trustworthy information. We also point to key challenges that remain. Ultimately, a range of empirical, technical, and theoretical advancements are needed to fully realize the promise of AI-mediated deliberation for enhancing citizen engagement and strengthening democratic deliberation.

cs.CY↗

arXiv2025

Democratic AI is Possible. The Democracy Levels Framework Shows How It Might Work

This position paper argues that effectively "democratizing AI" requires democratic governance and alignment of AI, and that this is particularly valuable for decisions with systemic societal impacts. Initial steps -- such as Meta's Community Forums and Anthropic's Collective Constitutional AI -- have illustrated a promising direction, where democratic processes could be used to meaningfully improve public involvement and trust in critical decisions. To more concretely explore what increasingly democratic AI might look like, we provide a "Democracy Levels" framework and associated tools that: (i) define milestones toward meaningfully democratic AI, which is also crucial for substantively pluralistic, human-centered, participatory, and public-interest AI, (ii) can help guide organizations seeking to increase the legitimacy of their decisions on difficult AI governance and alignment questions, and (iii) support the evaluation of such efforts.

cs.CY↗

arXiv2025

Scaling Human Judgment in Community Notes with LLMs

This paper argues for a new paradigm for Community Notes in the LLM era: an open ecosystem where both humans and LLMs can write notes, and the decision of which notes are helpful enough to show remains in the hands of humans. This approach can accelerate the delivery of notes, while maintaining trust and legitimacy through Community Notes' foundational principle: A community of diverse human raters collectively serve as the ultimate evaluator and arbiter of what is helpful. Further, the feedback from this diverse community can be used to improve LLMs' ability to produce accurate, unbiased, broadly helpful notes--what we term Reinforcement Learning from Community Feedback (RLCF). This becomes a two-way street: LLMs serve as an asset to humans--helping deliver context quickly and with minimal effort--while human feedback, in turn, enhances the performance of LLMs. This paper describes how such a system can work, its benefits, key new risks and challenges it introduces, and a research agenda to solve those challenges and realize the potential of this approach.

cs.CY↗

arXiv2025

Using Collective Dialogues and AI to Find Common Ground Between Israeli and Palestinian Peacebuilders

A growing body of work has shown that AI-assisted methods -- leveraging large language models, social choice methods, and collective dialogues -- can help navigate polarization and surface common ground in controlled lab settings. But what can these approaches contribute in real-world contexts? We present a case study applying these techniques to find common ground between Israeli and Palestinian peacebuilders in the period following October 7th, 2023. From April to July 2024 an iterative deliberative process combining LLMs, bridging-based ranking, and collective dialogues was conducted in partnership with the Alliance for Middle East Peace. Around 138 civil society peacebuilders participated including Israeli Jews, Palestinian citizens of Israel, and Palestinians from the West Bank and Gaza. The process resulted in a set of collective statements, including demands to world leaders, with at least 84% agreement from participants on each side. In this paper, we document the process, results, challenges, and important open questions.

cs.HC↗

arXiv2025

Language Agents as Digital Representatives in Collective Decision-Making

Consider the process of collective decision-making, in which a group of individuals interactively select a preferred outcome from among a universe of alternatives. In this context, "representation" is the activity of making an individual's preferences present in the process via participation by a proxy agent -- i.e. their "representative". To this end, learned models of human behavior have the potential to fill this role, with practical implications for multi-agent scenario studies and mechanism design. In this work, we investigate the possibility of training \textit{language agents} to behave in the capacity of representatives of human agents, appropriately expressing the preferences of those individuals whom they stand for. First, we formalize the setting of \textit{collective decision-making} -- as the episodic process of interaction between a group of agents and a decision mechanism. On this basis, we then formalize the problem of \textit{digital representation} -- as the simulation of an agent's behavior to yield equivalent outcomes from the mechanism. Finally, we conduct an empirical case study in the setting of \textit{consensus-finding} among diverse humans, and demonstrate the feasibility of fine-tuning large language models to act as digital representatives.

cs.LG↗

arXiv2024

Supernotes: Driving Consensus in Crowd-Sourced Fact-Checking

X's Community Notes, a crowd-sourced fact-checking system, allows users to annotate potentially misleading posts. Notes rated as helpful by a diverse set of users are prominently displayed below the original post. While demonstrably effective at reducing misinformation's impact when notes are displayed, there is an opportunity for notes to appear on many more posts: for 91% of posts where at least one note is proposed, no notes ultimately achieve sufficient support from diverse users to be shown on the platform. This motivates the development of Supernotes: AI-generated notes that synthesize information from several existing community notes and are written to foster consensus among a diverse set of users. Our framework uses an LLM to generate many diverse Supernote candidates from existing proposed notes. These candidates are then evaluated by a novel scoring model, trained on millions of historical Community Notes ratings, selecting candidates that are most likely to be rated helpful by a diverse set of users. To test our framework, we ran a human subjects experiment in which we asked participants to compare the Supernotes generated by our framework to the best existing community notes for 100 sample posts. We found that participants rated the Supernotes as significantly more helpful, and when asked to choose between the two, preferred the Supernotes 75.2% of the time. Participants also rated the Supernotes more favorably than the best existing notes on quality, clarity, coverage, context, and argumentativeness. Finally, in a follow-up experiment, we asked participants to compare the Supernotes against LLM-generated summaries and found that the participants rated the Supernotes significantly more helpful, demonstrating that both the LLM-based candidate generation and the consensus-driven scoring play crucial roles in creating notes that effectively build consensus among diverse users.

cs.SI↗

arXiv2022

Fine-tuning language models to find agreement among humans with diverse preferences

Recent work in large language modeling (LLMs) has used fine-tuning to align outputs with the preferences of a prototypical user. This work assumes that human preferences are static and homogeneous across individuals, so that aligning to a a single "generic" user will confer more general alignment. Here, we embrace the heterogeneity of human preferences to consider a different challenge: how might a machine help people with diverse views find agreement? We fine-tune a 70 billion parameter LLM to generate statements that maximize the expected approval for a group of people with potentially diverse opinions. Human participants provide written opinions on thousands of questions touching on moral and political issues (e.g., "should we raise taxes on the rich?"), and rate the LLM's generated candidate consensus statements for agreement and quality. A reward model is then trained to predict individual preferences, enabling it to quantify and rank consensus statements in terms of their appeal to the overall group, defined according to different aggregation (social welfare) functions. The model produces consensus statements that are preferred by human users over those from prompted LLMs (>70%) and significantly outperforms a tight fine-tuned baseline that lacks the final ranking step. Further, our best model's consensus statements are preferred over the best human-generated opinions (>65%). We find that when we silently constructed consensus statements from only a subset of group members, those who were excluded were more likely to dissent, revealing the sensitivity of the consensus to individual contributions. These results highlight the potential to use LLMs to help groups of humans align their values with one another.

cs.LG↗