While the success of large language models (LLMs) increases demand for
machine-generated text, current pay-per-token pricing schemes create a
misalignment of incentives known in economics as moral hazard: Text-generating
agents have strong incentive to cut costs by preferring a cheaper model over
the cutting-edge one, and this can be done "behind the scenes" since the agent
performs inference internally. In this work, we approach this issue from an
economic perspective, by proposing a pay-for-performance, contract-based
framework for incentivizing quality. We study a principal-agent game where the
agent generates text using costly inference, and the contract determines the
principal's payment for the text according to an automated quality evaluation.
Since standard contract theory is inapplicable when internal inference costs
are unknown, we introduce cost-robust contracts. As our main theoretical
contribution, we characterize optimal cost-robust contracts through a direct
correspondence to optimal composite hypothesis tests from statistics,
generalizing a result of Saig et al. (NeurIPS'23). We evaluate our framework
empirically by deriving contracts for a range of objectives and LLM evaluation
benchmarks, and find that cost-robust contracts sacrifice only a marginal
increase in objective value compared to their cost-aware counterparts.

当前的大型语言模型（LLMs）成功增加了对机器生成文本的需求，然而目前的按令牌计费模式存在一种经济学上称为道德风险的激励错位：生成文本的代理有强烈的动机通过偏好更便宜的模型来降低成本，并且这可以在内部执行推理的过程中完成。我们从经济学角度提出了一种以绩效为基础的以合同为基础的框架来解决这个问题。通过研究代理生成文本的贵重推理过程，并以自动质量评估来决定合同确定主体根据文本支付的金额，我们研究了一个主体 - 代理博弈。由于标准合同理论无法适用于不知道内部推理成本的情况，我们引入了成本 - 稳健合同。作为我们的主要理论贡献，我们通过与统计学中最优复合假设检验的直接对应关系，对成本 - 稳健合同进行了表征，推广了 Saig 等人的研究（NeurIPS'23）的结果。我们通过为一系列目标和 LLM 评估基准推导合同来实证评估我们的框架，并发现与成本意识的合同相比，成本 - 稳健合同只需要微不足道的目标价值增加。

通过统计合约激励高质量文本生成

Incentivizing Quality Text Generation via Statistical Contracts

In this paper, we tackle the issue of moral hazard within the realm of the
vehicular Metaverse. A pivotal facilitator of the vehicular Metaverse is the
effective orchestration of its market elements, primarily comprised of sensing
internet of things (SIoT) devices. These SIoT devices play a critical role by
furnishing the virtual service provider (VSP) with real-time sensing data,
allowing for the faithful replication of the physical environment within the
virtual realm. However, SIoT devices with intentional misbehavior can identify
a loophole in the system post-payment and proceeds to deliver falsified
content, which cause the whole vehicular Metaverse to collapse. To combat this
significant problem, we propose an incentive mechanism centered around a
reputation-based strategy. Specifically, the concept involves maintaining
reputation scores for participants based on their interactions with the VSP.
These scores are derived from feedback received by the VSP from Metaverse users
regarding the content delivered by the VSP and are managed using a subjective
logic model. Nevertheless, to prevent ``good" SIoT devices with false positive
ratings to leave the Metaverse market, we build a vanishing-like system of
previous ratings so that the VSP can make informed decisions based on the most
recent and accurate data available. Finally, we validate our proposed model
through extensive simulations. Our primary results show that our mechanism can
efficiently prevent malicious devices from starting their poisoning attacks. At
the same time, trustworthy SIoT devices that had a previous miss-classification
are not banned from the market.

在这篇论文中，我们通过提出一个以声誉为基础的策略，解决了在车载多元宇宙中的道德风险问题，其中的关键是通过感知物联网设备来实现交付真实的数据并防止恶意的攻击。

增强车辆元宇宙中的信任与安全：基于声誉的道德风险参与者机制

Enhancing Trust and Security in the Vehicular Metaverse: A  Reputation-Based Mechanism for Participants with Moral Hazard

Artificial Intelligence (AI) has received an increasing amount of attention
in multiple areas. The uncertainties and risks in AI-powered systems have
created reluctance in their wild adoption. As an economic solution to
compensate for potential damages, AI liability insurance is a promising market
to enhance the integration of AI into daily life. In this work, we use an
AI-powered E-diagnosis system as an example to study AI liability insurance. We
provide a quantitative risk assessment model with evidence-based numerical
analysis. We discuss the insurability criteria for AI technologies and suggest
necessary adjustments to accommodate the features of AI products. We show that
AI liability insurance can act as a regulatory mechanism to incentivize
compliant behaviors and serve as a certificate of high-quality AI systems.
Furthermore, we suggest premium adjustment to reflect the dynamic evolution of
the inherent uncertainty in AI. Moral hazard problems are discussed and
suggestions for AI liability insurance are provided.

本文提出了一种用于定量风险评估的模型，并讨论了人工智能责任保险的可行性和必要性，旨在促进人工智能的应用和规范。同时，我们建议通过动态调整保险费来反映人工智能固有不确定性的动态演化。本文还探讨了道德风险问题并提供了一些建议。