betaFeedback

WorksComputer Science

OmniMechanism Design for Human-AI Collaboration with Impressionable Minds

Abstract

As AI systems are increasingly deployed in high-stakes domains such as healthcare, effectively integrating human expertise with AI outputs is critical for reliable decision-making. A substantial body of theoretical work studies optimal mechanisms for human-AI collaboration (HAC), focusing on when to involve humans and how to combine their expertise with AI predictions. However, these approaches largely overlook a key factor highlighted by empirical evidence: human judgments are impressionable and can vary depending on how AI outputs are presented. For example, when AI systems defer decisions to humans, revealing or withholding the AI prediction can significantly influence human judgments. Moreover, existing work neglects deployment-time variability, where conditions, particularly the cost of eliciting human input, may differ substantially from those observed during training. To address these challenges, we introduce a rigorous and theoretically grounded framework for human-AI collaboration, termed collaborative omnimechanism design, based on a novel and computationally efficient contextual bandit variant of boosting for outcome indistinguishability. In particular, we reduce a calibration-type outcome indistinguishability condition to weighted calibration, providing a novel time-efficiency result, bypassing the high-dimension curse induced by complicated collaboration methods and outcome space. Our framework provides a principled characterization of human impressionability and introduces a unified optimization scheme that enables robust and adaptable decision-making across diverse query costs and evaluation objectives without retraining. We evaluate the framework in three HAC settings, examining adaptation to query costs and prediction objectives without retraining, and show that its induced policies perform comparably to or better than the evaluated baselines.

Introduction

Despite rapid advances in machine learning, human judgment remains essential in many high-stakes decision-making tasks. In domains such as healthcare, AI systems can provide accurate predictions, yet often lack access to the full context available to human experts. Conversely, human decision-makers can incorporate rich contextual information but may benefit from algorithmic guidance. This complementary structure has led to growing interest in human-AI collaboration (HAC) as a framework for combining the strengths of both [23, 9, 15, 6, 3, 1, 14, 4, 2].

A key but often overlooked aspect of human-AI collaboration is the impressionability of human decision-makers. In HAC settings, when an AI system defers a task to a human while providing a prediction or auxiliary signal, this information can influence the human’s perception of the task and alter their subsequent behavior and final decision. For example, Bondi et al. [ [9] show that, when deferring a task to a human, it is possible to improve human performance by informing the human of the deferral decision without revealing the AI prediction. Vodrahalli et al. [ [35] further demonstrate that manipulating the displayed confidence score of an AI system can alter subsequent human judgments. More recently, Bo et al. [ [8] reveal that, in collaboration with large language models, different reliance interventions can significantly affect both decision quality and the time required for users to engage with AI advice and answer questions. We use impressionability to mean that a human’s response distribution depends on how AI advice is presented. This impressionability creates a fundamental challenge: the presentation of AI outputs becomes a decision variable to optimize, while the resulting distribution of human responses depends on this choice. This dependence is reminiscent of performative prediction [31, 27, 26]. Despite growing empirical evidence of such effects [6, 9, 25, 7, 8, 11], many theoretical HAC models treat human responses as independent of the presentation of AI advice.

Another fundamental challenge in human–AI collaboration arises from deployment-time variability. In practice, the downstream environment in which a collaboration mechanism is used can differ significantly from the conditions assumed during training. In particular, the evaluation criteria and the cost of querying human expertise may vary across applications, time, and resource constraints [30, 28, 5, 10]. For example, in a clinical setting, the cost of consulting a specialist may vary depending on workload or urgency, while the evaluation criterion may shift between minimizing diagnostic error and prioritizing high-risk patients, depending on the application. Moreover, once a system is deployed, access to training data is often limited or unavailable, making it impractical to retrain the model for each new setting. Addressing this challenge requires frameworks that can adapt collaboration policies to varying deployment-time prediction objectives and human-query costs without retraining the learned outcome model. However, existing approaches typically optimize for a single objective and do not directly support such adaptation [30, 28, 34, 32].

Our contributions. To address the challenges of human impressionability and deployment-time variability, we introduce a unified theoretical framework for human-AI collaboration based on collaborative omnimechanism design. Our main contribution is the theoretical foundation of Collaborative Outcome Indistinguishability (COI), a loss-matching condition underlying a computationally tractable omniprediction framework tailored for human-AI collaboration with impressionable human minds. Formally, the COI condition ensures that expected losses under our learned predictive model closely match those under the true data-generating environment for the policies in a prescribed comparison class and the policies induced by the learned model across supported loss-cost pairs. This enables a central planner to select a policy whose expected loss is close to that of the best policy in the comparison class for each supported loss-cost pair, without retraining or requiring an explicit parametric model of impressionable human behavior. The core technical contributions of our work are as follows:

(a). Efficient COI learning via boosting. We extend the omniprediction framework to the HAC setting and propose an efficient boosting algorithm, COI-Boost, that achieves outcome indistinguishability using a reliable auditor. We establish finite-sample guarantees for both the auditor and the boosting procedure, and validate the effectiveness of this approach empirically. Further, we reduce Decision-OI auditing to weighted calibration and obtain a computational guarantee under the coverage and smoothing assumptions. We also derive a sample-complexity lower bound for offline omnimechanism learning.

(b). Capturing impressionable human behavior. We do not posit a parametric model of human responses or require pointwise recovery of their true distribution. Instead, we learn a surrogate outcome model whose expected losses agree with those of the environment for the relevant policies and objectives under standard coverage and complexity assumptions. This enables near-optimal decision-making without requiring a precise model of human behavior.

(c). Extensions to richer collaboration settings. We extend our framework to settings with multiple human experts and to feature-based methods, demonstrating its flexibility and applicability to more complex human-AI collaboration scenarios.

(d). Evaluation across decision regimes. We evaluate cost adaptation with human responses generated by a simulator fitted to real interaction data in binary classification, observed radiologist–AI interactions, and house-price prediction. Together, these experiments test whether a learned predictive model can support input-dependent collaboration policies across distinct decision regimes and adapt to different costs without retraining.

Setup

In this section, we formally introduce our mathematical models and desiderata.

Basic mechanism design with a single loss. For clarity, we first formulate collaboration for a fixed prediction loss and human-query cost, then extend the objective to a family of losses and costs.

As illustrated in Figure 1, for an input x∈Xx \in\mathcal{X}, a black-box AI model MM produces a prediction yM∈Yy_M \in\mathcal{Y}. In human-AI collaboration, given the context z=(x,yM)z = (x, y_M), the mechanism first decides whether to request human input using a deferral rule r:Z→{0,1}r : \mathcal{Z} \to\{0,1\}, where Z=X×Y\mathcal{Z} = \mathcal{X} \times\mathcal{Y}. This leads to two cases:

OmniMechanism setup and workflow diagram

Figure 1. OmniMechanism setup and workflow diagram.

When r(z)=0r(z) = 0, the process is fully automated, and we do not rely on any human input; the AI model’s prediction yMy_M serves as the final output.

When r(z)=1r(z) = 1, our mechanism presents the transformed (possibly augmented, curated, or reframed) AI output, which we also refer to as the signal, s=ϕ(z)∈Ss = \phi(z) \in\mathcal{S}, to a human decision-maker, resulting in a human output yH∈Yy_H \in\mathcal{Y}. Here, the function ϕ:Z→S\phi: \mathcal{Z} \to\mathcal{S} denotes the signal-presentation rule. Examples include the parameterized confidence score manipulation function [35] and different types of reliance intervention [8] mentioned in the introduction. Importantly, because the human decision-maker is influenced by the presented signal, different choices of the signal ss induce different distributions over the human output. Our mechanism then combines the AI model’s prediction and the human output through a collaboration function gg, yielding the final output g(yM,yH)∈Yg(y_M,y_H) \in\mathcal{Y}. In this process, we must choose both ϕ\phi and gg in order to optimize the quality of the final output. Let R\mathcal{R}, Φ\Phi, and G\mathcal{G} denote prescribed classes of deferral rules r:Z→{0,1}r : \mathcal{Z} \to\{0,1\}, signal-presentation rules ϕ:Z→S\phi: \mathcal{Z} \to\mathcal{S}, and collaboration functions g:Y×Y→Yg : \mathcal{Y} \times\mathcal{Y} \to\mathcal{Y}, respectively.

In both cases, the final output quality is evaluated against the ground truth ytrue∈Ytruey_{\mathrm{true}} \in\mathcal{Y}_{\mathrm{true}} corresponding to input xx. For notational convenience, we write the final prediction as yfinaly_{\mathrm{final}}, which is set to be yMy_M when r(z)=0r(z) = 0, and g(yM,yH)g(y_M,y_H) when r(z)=1r(z) = 1. We define H={(r,ϕ,g)∣r∈R,ϕ∈Φ,g∈G}\mathcal{H} = \{(r,\phi,g)|r \in\mathcal{R},\phi\in\Phi,g \in\mathcal{G}\}. Each triple h=(r,ϕ,g)∈Hh = (r,\phi,g) \in\mathcal{H} specifies a collaboration mechanism consisting of a deferral rule, a signal-presentation rule, and a collaboration function. Our objective must account for both the cost of human querying and the quality of the final prediction. Thus, given a fixed function pair ω=(l~,c)\omega= (\tilde{l},c), where loss l~\tilde{l} measures prediction error and cost function c(x)c(x) represents the cost of querying the human on input xx, we consider the following total loss at h∈Hh \in\mathcal{H}:

ℓω(z,yH,ytrue;h)=(1−r(z))l~(yM,ytrue)+r(z)[l~(g(yM,yH),ytrue)+c(x)].\ell^\omega(z,y_H,y_{\mathrm{true}};h) = (1-r(z))\tilde{l}(y_M,y_{\mathrm{true}}) + r(z)\left[\tilde{l}(g(y_M,y_H),y_{\mathrm{true}}) + c(x)\right].

Here, the expected loss depends on r,gr,g and most importantly, depends on the signal through the distribution of yHy_H. We also note that, in general, Ytrue\mathcal{Y}_{\mathrm{true}} may differ from Y\mathcal{Y}. For example, Ytrue={0,1}\mathcal{Y}_{\mathrm{true}} = \{0,1\} may represent binary ground-truth labels, while Y=[0,1]\mathcal{Y} = [0,1] represents probability-valued predictions.

Example 2.1 (Illustration of Notations). Given a patient’s medical image and clinical records xx, a black-box AI model first produces a diagnosis prediction yMy_M. The system then decides whether to rely directly on the AI prediction or defer the case to a physician. If the task is deferred, the physician is shown an additional signal s=ϕ(z)s = \phi(z) derived from the AI output, such as a displayed confidence score or a reliability warning. For example, presenting a highly confident AI prediction may increase the physician’s reliance on the AI recommendation, while a cautionary intervention may encourage more careful review. Because human judgments are impressionable, different presentations can lead to different physician decisions. The final diagnosis is then produced by combining the AI prediction and the physician’s assessment through a collaboration function gg [4].

From the previous discussion, we see that, to design an (optimal) mechanism for human-AI collaboration with impressionable human decision-makers, given a black-box AI model, we need to choose a deferral function r∈Rr \in\mathcal{R}, a signal-presentation rule ϕ∈Φ\phi\in\Phi, and a collaboration function g∈Gg \in\mathcal{G}. In a typical supervised learning framework with a fixed prediction loss l~\tilde{l} and a cost function cc, the goal will be achieving

min⁡h∈HEz,yH,ytrueℓω(z,yH,ytrue;h),\min_{h \in\mathcal{H}} \mathbb{E}_{z,y_H,y_{\mathrm{true}}}\ell^\omega(z,y_H,y_{\mathrm{true}};h),

The expectation is taken under the distribution induced by the candidate signal presentation rule ϕ\phi, whose effect on human responses is unknown, while the context distribution is held fixed throughout.

Our ultimate goal: omnimechanism design for a loss family. As motivated in the introduction, in human-AI collaboration, in many cases, neither the prediction loss l~\tilde{l} nor the querying cost cc is fixed or known a priori, as both depend on the downstream task, resource constraints, and deployment context. To address this, we adopt an omniprediction perspective [18, 19, 13] (Dwork et al. 2021) and seek a mechanism design that simultaneously performs well across a class of losses l~\tilde{l} and cost functions cc without retraining, ensuring robustness and broad applicability without needing to tailor the model to a specific downstream objective.

Let us consider a family of prediction losses L\mathcal{L} and a family of query-cost functions C\mathcal{C}, and let Ω=L×C\Omega= \mathcal{L} \times\mathcal{C}. We further denote Δ(T)\Delta(T) as the set of probability distributions over a set TT. Inspired by the omniprediction theory [10, 18], we learn a base predictive distribution p:Z×{0,1}×S×G→Δ(Y×Ytrue)p : \mathcal{Z} \times\{0,1\} \times\mathcal{S} \times\mathcal{G} \to\Delta(\mathcal{Y} \times\mathcal{Y}_{\mathrm{true}}) modeling the joint distribution of the resulting final prediction and the ground-truth label, such that, for any pair ω=(l~,c)∈Ω\omega= (\tilde{l},c) \in\Omega of loss and cost, there is a data-free post-processing map kωk^\omega satisfying [kω(p)](z)=(rω(z),ϕω(z),gω(z,⋅,⋅))[k^\omega(p)](z) = (r^\omega(z),\phi^\omega(z),g^\omega(z,\cdot,\cdot)), which maps from Z\mathcal{Z} to {0,1}×S×G\{0,1\} \times\mathcal{S} \times\mathcal{G} (i.e., a tuple specifying a deferral decision, signaling strategy, and a zz-dependent collaboration function), thereby defining

ℓω(z,yH,ytrue;kω(p))=(1−rω(z))l~(yM,ytrue)+rω(z)[l~(gω(z,yM,yH),ytrue)+c(x)],\ell^\omega(z,y_H,y_{\mathrm{true}};k^\omega(p)) = (1-r^\omega(z))\tilde{l}(y_M,y_{\mathrm{true}}) + r^\omega(z)\left[\tilde{l}(g^\omega(z,y_M,y_H),y_{\mathrm{true}}) + c(x)\right],

where human output yHy_H implicitly depends on ϕω(z)\phi^\omega(z). We highlight a notational distinction: while any base collaboration function g∈Gg \in\mathcal{G} operates exclusively on the predictions, mapping Y×Y\mathcal{Y} \times\mathcal{Y} to Y\mathcal{Y}, our post-processing map extracts the collaboration function rule pointwise using the base predictive distribution p(z)p(z) for each state zz. Therefore, for each context zz, the selected collaboration function gω(z,⋅,⋅)g^\omega(z,\cdot,\cdot) belongs to G\mathcal{G} while the selected function may vary with zz.

Loosely speaking, for any ϵ>0\epsilon> 0, we say that pp is an (Ω,H,ϵ)(\Omega,\mathcal{H},\epsilon)-collaborative omnimechanism if for any loss-cost pair ω∈Ω\omega\in\Omega,

Ez,yH,ytrueℓω(z,yH,ytrue;kω(p))≤min⁡h∈HEz,yH,ytrueℓω(z,yH,ytrue;h)+ϵ.\mathbb{E}_{z,y_H,y_{\mathrm{true}}}\ell^\omega(z,y_H,y_{\mathrm{true}};k^\omega(p)) \leq\min_{h \in\mathcal{H}}\mathbb{E}_{z,y_H,y_{\mathrm{true}}}\ell^\omega(z,y_H,y_{\mathrm{true}};h) + \epsilon.

The guarantee compares the induced policy with the best mechanism in H\mathcal{H}. The induced policy can be more expressive because it may choose a different aggregation rule for each context. We remark that the left hand side expectation uses the response distribution induced by the signal-presentation rule ϕω\phi^\omega while each right-hand expectation uses the distribution induced by the corresponding prescribed base policy’s signal-presentation rule.

Our goal is to develop an omnimechanism that can simultaneously achieve approximate optimality across different choices of l~\tilde{l} and cc through simple, data-free post-processing, without requiring retraining. For clarity, this section describes deterministic post-processing. § 3.1 introduces a smoothed randomized version and Theorem 4.1 quantifies the additional approximation error. Building on the omniprediction framework [Gopalan et al. 2021; Kim and [20] (Perdomo 2022), we will develop a sample-efficient boosting algorithm for the smoothed collaborative omnimechanism construction.

Methodology

To formally tackle the omnimechanism design objective, we must address the fundamental challenge of our setting: the impressionability of human behavior. Because the human output yHy_{\mathrm{H}} shifts based on the signal ss they are shown, we recast our framework in the language of contextual bandits. Building on outcome indistinguishability [12, 24], we employ a boosting algorithm to construct the base predictive distribution pp, thereby validating the omnimechanism. We then show how to post-process pp in a data-free manner to obtain near-optimality guarantees.

Policy Learning and State-Action Space

We treat omnimechanism design as a policy learning problem. We first learn an outcome model from offline human-AI interaction data. At deployment, we use this model to evaluate the expected total losses of candidate actions under a supplied prediction loss and human-query cost, and obtain a collaboration policy by post-processing without retraining. We express this procedure in the language of contextual bandits, with the policies πh\pi_h induced by mechanisms h∈Hh \in\mathcal{H} serving as the comparison class.

References

  1. [1]Nikhil Agarwal, Alex Moehring, Pranav Rajpurkar, and Tobias Salz. Combining human expertise with artificial intelligence: Experimental evidence from radiology. Technical report, National Bureau of Economic Research, 2023.DOI
  2. [2]Nikhil Agarwal, Alex Moehring, and Alexander Wolitzky. Designing human-ai collaboration: A sufficient-statistic approach. Technical report, National Bureau of Economic Research, 2025.DOI
  3. [3]Rohan Alur, Loren Laine, Darrick Li, Manish Raghavan, Devavrat Shah, and Dennis Shung. Auditing for human expertise. Advances in Neural Information Processing Systems, 36:79439–79468, 2023.arxiv.org/abs/2306.01646
  4. [4]Rohan Alur, Manish Raghavan, and Devavrat Shah. Human expertise in algorithmic prediction. Advances in Neural Information Processing Systems, 37:138088–138129, 2024.arxiv.org/abs/2402.00793
  5. [5]Jean V Alves, Diogo Leitão, Sérgio Jesus, Marco OP Sampaio, Javier Liébana, Pedro Saleiro, Mário AT Figueiredo, and Pedro Bizarro. Cost-sensitive learning to defer to multiple experts with workload constraints. arXiv preprint arXiv:2403.06906, 2024.
  6. [6]Gagan Bansal, Besmira Nushi, Ece Kamar, Walter S Lasecki, Daniel S Weld, and Eric Horvitz. Beyond accuracy: The role of mental models in human-ai team performance. In Proceedings of the AAAI conference on human computation and crowdsourcing, volume 7, pages 2–11, 2019.
  7. [7]Tina Behzad, Nikolos Gurney, Ning Wang, and David V Pynadath. Beyond predictions: A study of ai strength and weakness transparency communication on human-ai collaboration. In International Conference on Human-Computer Interaction, pages 105–131. Springer, 2025.
  8. [8]Jessica Y Bo, Sophia Wan, and Ashton Anderson. To rely or not to rely? evaluating interventions for appropriate reliance on large language models. In Proceedings of the 2025 CHI Conference on Human Factors in Computing Systems, pages 1–23, 2025.
  9. [9]Elizabeth Bondi, Raphael Koster, Hannah Sheahan, Martin Chadwick, Yoram Bachrach, Taylan Cemgil, Ulrich Paquet, and Krishnamurthy Dvijotham. Role of human-ai interaction in selective prediction. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 36, pages 5286–5294, 2022.arxiv.org/abs/2112.06751
  10. [10]Sílvia Casacuberta and Varun Kanade. Selective omniprediction and fair abstention. In The Thirty-ninth Annual Conference on Neural Information Processing Systems, 2025. URL https://openreview.net/forum?id=BoYGLpNXZd.DOI
  11. [11]Paul Duetting, Safwan Hossain, Tao Lin, Renato Paes Leme, Sai Srivatsa Ravindranath, Haifeng Xu, and Song Zuo. Information design with large language models. arXiv preprint arXiv:2509.25565, 2025.arxiv.org/abs/2509.25565
  12. [12]Cynthia Dwork, Michael P. Kim, Omer Reingold, Guy N. Rothblum, and Gal Yona. Outcome indistinguishability. In Proceedings of the 53rd Annual ACM SIGACT Symposium on Theory of Computing, STOC 2021, page 1095–1108, New York, NY, USA, 2021. Association for Computing Machinery. ISBN 9781450380539. doi: 10.1145/3406325.3451064. URL https://doi.org/10.1145/3406325.3451064.
  13. [13]Cynthia Dwork, Michael P Kim, Omer Reingold, Guy N Rothblum, and Gal Yona. Beyond bernoulli: Generating random outcomes that cannot be distinguished from nature. In International Conference on Algorithmic Learning Theory, pages 342–380. PMLR, 2022.
  14. [14]Xueyang Feng, Zhi-Yuan Chen, Yujia Qin, Yankai Lin, Xu Chen, Zhiyuan Liu, and Ji-Rong Wen. Large language model-based human-agent collaboration for complex task solving. In Findings of the Association for Computational Linguistics: EMNLP 2024, pages 1336–1357, 2024.
  15. [15]Yonatan Geifman and Ran El-Yaniv. Selective classification for deep neural networks. Advances in neural information processing systems, 30, 2017.arxiv.org/abs/1705.08500
  16. [16]Surbhi Goel, Varun Kanade, Adam Klivans, and Justin Thaler. Reliably learning the relu in polynomial time. In Conference on Learning Theory, pages 1004–1042. PMLR, 2017.arxiv.org/abs/1611.10258
  17. [17]Paul Goldberg and Mark Jerrum. Bounding the vapnik-chervonenkis dimension of concept classes parameterized by real numbers. In Proceedings of the sixth annual conference on Computational learning theory, pages 361–369, 1993.
  18. [18]Parikshit Gopalan, Adam Tauman Kalai, Omer Reingold, Vatsal Sharan, and Udi Wieder. Omnipredictors. arXiv preprint arXiv:2109.05389, 2021.arxiv.org/abs/2109.05389
  19. [19]Parikshit Gopalan, Lunjia Hu, Michael P Kim, Omer Reingold, and Udi Wieder. Loss minimization through the lens of outcome indistinguishability. arXiv preprint arXiv:2210.08649, 2022.arxiv.org/abs/2210.08649
  20. [20]Parikshit Gopalan, Lunjia Hu, and Guy N Rothblum. On computationally efficient multi-class calibration. In The Thirty Seventh Annual Conference on Learning Theory, pages 1983–2026. PMLR, 2024.arxiv.org/abs/2402.07821
  21. [21]Ziyang Guo, Yifan Wu, Jason D Hartline, and Jessica Hullman. A decision theoretic framework for measuring ai reliance. In Proceedings of the 2024 ACM conference on fairness, accountability, and transparency, pages 221–236, 2024.arxiv.org/abs/2401.15356
  22. [22]Ziyang Guo, Yifan Wu, Jason Hartline, and Jessica Hullman. The value of information in human-ai decision-making. In The Fourteenth International Conference on Learning Representations, 2025.
  23. [23]Maia Jacobs, Melanie F Pradier, Thomas H McCoy Jr, Roy H Perlis, Finale Doshi-Velez, and Krzysztof Z Gajos. How machine-learning recommendations influence clinician treatment selections: the example of antidepressant selection. Translational psychiatry, 11(1):108, 2021.DOI
  24. [24]Michael P Kim and Juan C Perdomo. Making decisions under outcome performativity. arXiv preprint arXiv:2210.01745, 2022.arxiv.org/abs/2210.01745
  25. [25]Jingshu Li, Yitian Yang, Renwen Zhang, Q Vera Liao, Tianqi Song, Zhengtao Xu, and Yi-chieh Lee. Understanding the effects of miscalibrated ai confidence on user trust, reliance, and decision efficacy. arXiv preprint arXiv:2402.07632, 2024.arxiv.org/abs/2402.07632
  26. [26]Xiang Li, Yunai Li, Huiying Zhong, Lihua Lei, and Zhun Deng. Statistical inference under performativity. arXiv preprint arXiv:2505.18493, 2025.arxiv.org/abs/2505.18493
  27. [27]Licong Lin and Tijana Zrnic. Plug-in performative optimization. arXiv preprint arXiv:2305.18728, 2023.arxiv.org/abs/2305.18728
  28. [28]Anqi Mao, Christopher Mohri, Mehryar Mohri, and Yutao Zhong. Two-stage learning to defer with multiple experts. Advances in neural information processing systems, 36:3578–3606, 2023.DOI
  29. [29]Alex Moehring, Manasi Kutwal, Ray Huang, Oishi Banerjee, Adam Jacobi, Corey Eber, Dexter Mendoza, Mike Chung, Etan Dayan, Yogesh Gupta, et al. A dataset for understanding radiologist-artificial intelligence collaboration. Scientific data, 12(1):739, 2025.DOI
  30. [30]Hussein Mozannar and David Sontag. Consistent estimators for learning to defer to an expert. In International conference on machine learning, pages 7076–7087. PMLR, 2020.arxiv.org/abs/2006.01862
  31. [31]Juan Perdomo, Tijana Zrnic, Celestine Mendler-Dünner, and Moritz Hardt. Performative prediction. In International Conference on Machine Learning, pages 7599–7609. PMLR, 2020.arxiv.org/abs/2002.06673
  32. [32]Nilesh Ramgolam. Learning to defer to a population with limited demonstrations. In 2025 International Conference on Digital Image Computing: Techniques and Applications (DICTA), pages 1–7. IEEE, 2025.arxiv.org/abs/2510.19351
  33. [33]Andrew Sellergren, Sahar Kazemzadeh, Tiam Jaroensri, Atilla Kiraly, Madeleine Traverse, Timo Kohlberger, Shawn Xu, Fayaz Jamil, Cían Hughes, Charles Lau, et al. Medgemma technical report. arXiv preprint arXiv:2507.05201, 2025.DOI
  34. [34]Rajeev Verma, Daniel Barrejon, and Eric Nalisnick. Learning to defer to multiple experts: Consistent surrogate losses, confidence calibration, and conformal ensembles. In International Conference on Artificial Intelligence and Statistics, pages 11415–11434. PMLR, 2023.arxiv.org/abs/2210.16955
  35. [35]Kailas Vodrahalli, Tobias Gerstenberg, and James Y Zou. Uncalibrated models can improve human-ai collaboration. Advances in Neural Information Processing Systems, 35:40004–40016, 2022.arxiv.org/abs/2202.05983

Paper details

vlnc:W_7uwbnxde
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 10

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Computer Science

Not recorded for this paper

View statement
Looking for related papers…
Contents