arXiv subjects

Aleksandrs Slivkins

Publications and source records attributed to Aleksandrs Slivkins.

At least 19 recordsLinked to original sources

arXiv2026

The Evolution of Digital Search: From Blue Links to Delegated Decision-Making

Digital search is undergoing a fundamental transformation from a human-driven process of discovery to an agent-mediated system of delegated decision-making. In the traditional model of digital search, users translate intent into keyword-based queries, evaluate ranked lists of links, and execute decisions outside the search interface. In an AI-native world, users express goals in natural language, agents interpret these intentions, and outcomes are returned as recommendations or executed decisions. This shift moves search from a link-based user interface to an embedded system component, with implications for transparency, competition, and monetization. The resulting system design problem raises key questions about information quality and access, trust, incentive alignment, and market structure. Early evidence from experimental agent-mediated marketplaces and economic theory suggests that small design choices, such as how stakeholders access information, how options are surfaced, and how actions are executed, have first-order effects on efficiency, competition, and the welfare of consumers and firms. We propose that the future of search will be determined not by incremental improvements in ranking algorithms and natural-language interfaces, but by the design of open, transparent, and competitive agentic systems that govern how decisions are made and how markets operate, highlighting a set of grand challenges at the intersection of AI, economics, and system design.

econ.GN↗

arXiv2026

Bandit Social Learning with Exploration Episodes

We study a stylized social learning dynamics where self-interested agents collectively follow a simple multi-armed bandit protocol. Each agent controls an ``episode": a short sequence of consecutive decisions. Motivating applications include users repeatedly interacting with an AI, or repeatedly shopping at a marketplace. While agents are incentivized to explore within their respective episodes, we show that the aggregate exploration fails: e.g., its Bayesian regret grows linearly over time. In fact, such failure is a (very) typical case, not just a worst-case scenario. This conclusion persists even if an agent's per-episode utility is some fixed function of the per-round outcomes: e.g., $\min$ or $\max$, not just the sum. Thus, externally driven exploration is needed even when some amount of exploration happens organically.

cs.GT↗

arXiv2026

Should You Use Your Large Language Model to Explore or Exploit?

We evaluate the ability of the current generation of large language models (LLMs) to help a decision-making agent facing an exploration-exploitation tradeoff. While previous work has largely study the ability of LLMs to solve combined exploration-exploitation tasks, we take a more systematic approach and use LLMs to explore and exploit in silos in various (contextual) bandit tasks. We find that reasoning models show the most promise for solving exploitation tasks, although they are still too expensive or too slow to be used in many practical settings. Motivated by this, we study tool use and in-context summarization using non-reasoning models. We find that these mitigations may be used to substantially improve performance on medium-difficulty tasks, however even then, all LLMs we study perform worse than a simple linear regression, even in non-linear settings. On the other hand, we find that LLMs do help at exploring large action spaces with inherent semantics, by suggesting suitable candidates to explore.

cs.LG↗

arXiv2026

From Augmentation to Reconstruction: Guiding the AI Disruption to the Good Place

Artificial intelligence feels omnipresent, yet the disruption many expect has not fully arrived. The main reason is not model capability, nor even the tools built to harness those models. Rather, most organizations are still using AI to accelerate workflows designed for a pre-AI world. We offer a three-stage lens: Augmentation, Automation, and Reconstruction, and argue that the most consequential disruption resides in the third stage where workflows and markets are rebuilt around delegation, machine-to-machine interaction, continuous monitoring, and auditable constraints. Achieving this system-level transformation takes time: it requires trust and accountability infrastructure, machine-legible and interoperable data and interfaces, the design and adoption of these new workflows, and economic incentives that favor reconstruction rather than local optimization: the complementary investments that produce the familiar "productivity J-curve" of general-purpose technologies. We illustrate this transition through examples in consumer markets, education, news, and coding. Finally, we emphasize a normative point: the agentic future is not predetermined. Leaders must both skate to where the puck is going and actively steer it toward a good place, ensuring innovation delivers welfare gains felt by businesses and consumers around the world.

econ.GN↗

arXiv2026

Incentivizing Exploration with Selective Data Disclosure

We propose and design recommendation systems that incentivize efficient exploration. Agents arrive sequentially, choose actions and receive rewards, drawn from fixed but unknown action-specific distributions. The recommendation system presents each agent with actions and rewards from a subsequence of past agents, chosen ex ante. Thus, the agents engage in sequential social learning, moderated by these subsequences. We asymptotically attain optimal regret rate for exploration, using a flexible frequentist behavioral model and mitigating rationality and commitment assumptions inherent in prior work. We suggest three components of effective recommendation systems: independent focus groups, group aggregators, and interlaced information structures.

cs.GT↗

arXiv2026

Agentic Markets: Equilibrium Effects of Improving Consumer Search

Motivated by agentic markets -- two-sided markets in which consumers and businesses are assisted by AI tools that facilitate consumers' search -- we study the impact of improved search technology on learning and welfare in markets. We put forth a model where consumers engage in costly search to acquire signals of product fit prior to purchase. The market tracks indications of fit for searched products and indications of quality for chosen products, thereby guiding searches. We characterize the long-run steady-state of the resulting dynamics as well as the impact of improving search technology. We find cheaper search improves learning and consumer surplus, whereas more informative search can degrade both unless the market learns as much as consumers about the products by, for example, ``reading the transcripts'' of agentic conversations. Finally, we consider the impact of search improvements on how businesses set prices. At equilibrium prices in symmetric markets, consumer surplus is improved by cheaper search but may be decreased by more informative search, due to weakened inter-business competition.

cs.GT↗

arXiv2026

Exploration and Incentivizing Participation in Randomized Trials

Participation incentives is a well-known issue inhibiting randomized controlled trials (RCTs) in medicine, as well as a potential cause of user dissatisfaction for RCTs in online platforms. We frame this issue as a non-standard exploration-exploitation tradeoff: an RCT would like to explore as uniformly as possible, whereas each "agent" (a patient or a user) prefers "exploitation", i.e., treatments that seem best. We incentivize participation by leveraging information asymmetry between the trial and the agents. We measure statistical performance via worst-case estimation error under adversarially generated outcomes, a standard objective for RCTs. We obtain a near-optimal solution in terms of this objective: an incentive-compatible mechanism with a particular guarantee, and a nearly matching impossibility result for any incentive-compatible mechanism. We consider three model variants: homogeneous agents (of the same "type" comprising beliefs and preferences), heterogeneous agents, and an extension that leverages estimated type frequencies to mitigate the influence of rare-but-difficult agent types.

cs.GT↗

arXiv2026

Budget Pacing in Repeated Auctions: Regret and Efficiency without Convergence

We study the aggregate welfare and individual regret guarantees of dynamic \emph{pacing algorithms} in the context of repeated auctions with budgets. Such algorithms are commonly used as bidding agents in Internet advertising platforms, adaptively learning to shade bids by a tunable linear multiplier in order to match a specified budget. We show that when agents simultaneously apply a natural form of gradient-based pacing, the liquid welfare obtained over the course of the learning dynamics is at least half the optimal expected liquid welfare obtainable by any allocation rule. Crucially, this result holds \emph{without requiring convergence of the dynamics}, allowing us to circumvent known complexity-theoretic obstacles of finding equilibria. This result is also robust to the correlation structure between agent valuations and holds for any \emph{core auction}, a broad class of auctions that includes first-price, second-price, and generalized second-price auctions as special cases. For individual guarantees, we further show such pacing algorithms enjoy \emph{dynamic regret} bounds for individual utility- and value-maximization, with respect to the sequence of budget-pacing bids, for any auction satisfying a monotone bang-for-buck property. To complement our theoretical findings, we provide semi-synthetic numerical simulations based on auction data from the Bing Advertising platform.

cs.GT↗

arXiv2025

Greedy Algorithm for Structured Bandits: A Sharp Characterization of Asymptotic Success / Failure

We study the greedy (exploitation-only) algorithm in bandit problems with a known reward structure. We allow arbitrary finite reward structures, while prior work focused on a few specific ones. We fully characterize when the greedy algorithm asymptotically succeeds or fails, in the sense of sublinear vs. linear regret as a function of time. Our characterization identifies a partial identifiability property of the problem instance as the necessary and sufficient condition for the asymptotic success. Notably, once this property holds, the problem becomes easy -- any algorithm will succeed (in the same sense as above), provided it satisfies a mild non-degeneracy condition. Our characterization extends to contextual bandits and interactive decision-making with arbitrary feedback. Examples demonstrating broad applicability and extensions to infinite reward structures are provided.

cs.LG↗

arXiv2025

Magentic Marketplace: An Open-Source Environment for Studying Agentic Markets

As LLM agents advance, they are increasingly mediating economic decisions, ranging from product discovery to transactions, on behalf of users. Such applications promise benefits but also raise many questions about agent accountability and value for users. Addressing these questions requires understanding how agents behave in realistic market conditions. However, previous research has largely evaluated agents in constrained settings, such as single-task marketplaces (e.g., negotiation) or structured two-agent interactions. Real-world markets are fundamentally different: they require agents to handle diverse economic activities and coordinate within large, dynamic ecosystems where multiple agents with opaque behaviors may engage in open-ended dialogues. To bridge this gap, we investigate two-sided agentic marketplaces where Assistant agents represent consumers and Service agents represent competing businesses. To study these interactions safely, we develop Magentic-Marketplace -- a simulated environment where Assistants and Services can operate. This environment enables us to study key market dynamics: the utility agents achieve, behavioral biases, vulnerability to manipulation, and how search mechanisms shape market outcomes. Our experiments show that frontier models can approach optimal welfare -- but only under ideal search conditions. Performance degrades sharply with scale, and all models exhibit severe first-proposal bias, creating 10-30x advantages for response speed over quality. These findings reveal how behaviors emerge across market conditions, informing the design of fair and efficient agentic marketplaces.

cs.MA↗

arXiv2025

Adaptive Discretization against an Adversary: Lipschitz bandits, Dynamic Pricing, and Auction Tuning

Lipschitz bandits is a prominent version of multi-armed bandits that studies large, structured action spaces such as the $[0,1]$ interval, where similar actions are guaranteed to have similar rewards. A central theme here is the adaptive discretization of the action space, which gradually ``zooms in'' on the more promising regions thereof. The goal is to take advantage of ``nicer'' problem instances, while retaining near-optimal worst-case performance. While the stochastic version of the problem is well-understood, the general version with adversarial rewards is not. We provide the first algorithm (\emph{Adversarial Zooming}) for adaptive discretization in the adversarial version, and derive instance-dependent regret bounds. In particular, we recover the worst-case optimal regret bound for the adversarial version, and the instance-dependent regret bound for the stochastic version. We apply our algorithm to several fundamental applications -- including dynamic pricing and auction reserve tuning -- all under adversarial reward models. While these domains often violate Lipschitzness, our analysis only requires a weaker version thereof, allowing for meaningful regret bounds without additional smoothness assumptions. Notably, we extend our results to multi-product dynamic pricing with non-smooth reward structures, a setting which does not even satisfy one-sided Lipschitzness.

cs.LG↗

arXiv2025

The Agentic Economy

Generative AI has transformed human-computer interaction by enabling natural language interfaces and the emergence of autonomous agents capable of acting on users' behalf. While early applications have improved individual productivity, these gains have largely been confined to predefined tasks within existing workflows. We argue that the more profound economic impact lies in reducing communication frictions between consumers and businesses. This shift could reorganize markets, redistribute power, and catalyze the creation of new products and services. We explore the implications of an agentic economy, where assistant agents act on behalf of consumers and service agents represent businesses, interacting programmatically to facilitate transactions. A key distinction we draw is between unscripted interactions -- enabled by technical advances in natural language and protocol design -- and unrestricted interactions, which depend on market structures and governance. We examine the current limitations of siloed and end-to-end agents, and explore future scenarios shaped by technical standards and market dynamics. These include the potential tension between agentic walled gardens and an open web of agents, implications for advertising and discovery, the evolution of micro-transactions, and the unbundling and rebundling of digital goods. Ultimately, we argue that the architecture of agentic communication will determine the extent to which generative AI democratizes access to economic opportunity.

cs.CY↗

arXiv2025

Bandit Social Learning: Exploration under Myopic Behavior

We study social learning dynamics motivated by reviews on online platforms. The agents collectively follow a simple multi-armed bandit protocol, but each agent acts myopically, without regards to exploration. We allow the greedy (exploitation-only) algorithm, as well as a wide range of behavioral biases. Specifically, we allow myopic behaviors that are consistent with (parameterized) confidence intervals for the arms' expected rewards. We derive stark learning failures for any such behavior, and provide matching positive results. The learning-failure results extend to Bayesian agents and Bayesian bandit environments. In particular, we obtain general, quantitatively strong results on failure of the greedy bandit algorithm, both for ``frequentist" and ``Bayesian" versions. Failure results known previously are quantitatively weak, and either trivial or very specialized. Thus, we provide a theoretical foundation for designing non-trivial bandit algorithms, \ie algorithms that intentionally explore, which has been missing from the literature. Our general behavioral model can be interpreted as agents' optimism or pessimism. The matching positive results entail a maximal allowed amount of optimism. Moreover, we find that no amount of pessimism helps against the learning failures, whereas even a small-but-constant fraction of extreme optimists avoids the failures and leads to near-optimal regret rates.

cs.GT↗

arXiv2025

Robust and Performance Incentivizing Algorithms for Multi-Armed Bandits with Strategic Agents

Motivated by applications such as online labor markets we consider a variant of the stochastic multi-armed bandit problem where we have a collection of arms representing strategic agents with different performance characteristics. The platform (principal) chooses an agent in each round to complete a task. Unlike the standard setting, when an arm is pulled it can modify its reward by absorbing it or improving it at the expense of a higher cost. The principle has to solve a mechanism design problem to incentivize the arms to give their best performance. However, since even with an effective mechanism agents may still deviate from rational behavior, the principal wants a robust algorithm that also gives a non-vacuous guarantee on the total accumulated rewards under non-equilibrium behavior. In this paper, we introduce a class of bandit algorithms that meet the two objectives of performance incentivization and robustness simultaneously. We do this by identifying a collection of intuitive properties that a bandit algorithm has to satisfy to achieve these objectives. Finally, we show that settings where the principal has no information about the arms' performance characteristics can be handled by combining ideas from second price auctions with our algorithms.

cs.GT↗

arXiv2025

Algorithmic Persuasion Through Simulation

We study a Bayesian persuasion game where a sender wants to persuade a receiver to take a binary action, such as purchasing a product. The sender is informed about the (real-valued) state of the world, such as the quality of the product, but only has limited information about the receiver's beliefs and utilities. Motivated by customer surveys, user studies, and recent advances in AI, we allow the sender to learn more about the receiver by querying an oracle that simulates the receiver's behavior. After a fixed number of queries, the sender commits to a messaging policy and the receiver takes the action that maximizes her expected utility given the message she receives. We characterize the sender's optimal messaging policy given any distribution over receiver types. We then design a polynomial-time querying algorithm that optimizes the sender's expected utility in this game. We also consider approximate oracles, more general query structures, and costly queries.

cs.GT↗

arXiv2024

Autobidders with Budget and ROI Constraints: Efficiency, Regret, and Pacing Dynamics

We study a game between autobidding algorithms that compete in an online advertising platform. Each autobidder is tasked with maximizing its advertiser's total value over multiple rounds of a repeated auction, subject to budget and return-on-investment constraints. We propose a gradient-based learning algorithm that is guaranteed to satisfy all constraints and achieves vanishing individual regret. Our algorithm uses only bandit feedback and can be used with the first- or second-price auction, as well as with any "intermediate" auction format. Our main result is that when these autobidders play against each other, the resulting expected liquid welfare over all rounds is at least half of the expected optimal liquid welfare achieved by any allocation. This holds whether or not the bidding dynamics converges to an equilibrium.

cs.GT↗

arXiv2024

Contextual Bandits with Packing and Covering Constraints: A Modular Lagrangian Approach via Regression

We consider contextual bandits with linear constraints (CBwLC), a variant of contextual bandits in which the algorithm consumes multiple resources subject to linear constraints on total consumption. This problem generalizes contextual bandits with knapsacks (CBwK), allowing for packing and covering constraints, as well as positive and negative resource consumption. We provide the first algorithm for CBwLC (or CBwK) that is based on regression oracles. The algorithm is simple, computationally efficient, and statistically optimal under mild assumptions. Further, we provide the first vanishing-regret guarantees for CBwLC (or CBwK) that extend beyond the stochastic environment. We side-step strong impossibility results from prior work by identifying a weaker (and, arguably, fairer) benchmark to compare against. Our algorithm builds on LagrangeBwK (Immorlica et al., FOCS 2019), a Lagrangian-based technique for CBwK, and SquareCB (Foster and Rakhlin, ICML 2020), a regression-based technique for contextual bandits. Our analysis leverages the inherent modularity of both techniques.

cs.LG↗

arXiv2024

Can large language models explore in-context?

We investigate the extent to which contemporary Large Language Models (LLMs) can engage in exploration, a core capability in reinforcement learning and decision making. We focus on native performance of existing LLMs, without training interventions. We deploy LLMs as agents in simple multi-armed bandit environments, specifying the environment description and interaction history entirely in-context, i.e., within the LLM prompt. We experiment with GPT-3.5, GPT-4, and Llama2, using a variety of prompt designs, and find that the models do not robustly engage in exploration without substantial interventions: i) Across all of our experiments, only one configuration resulted in satisfactory exploratory behavior: GPT-4 with chain-of-thought reasoning and an externally summarized interaction history, presented as sufficient statistics; ii) All other configurations did not result in robust exploratory behavior, including those with chain-of-thought reasoning but unsummarized history. Although these findings can be interpreted positively, they suggest that external summarization -- which may not be possible in more complex settings -- is important for obtaining desirable behavior from LLM agents. We conclude that non-trivial algorithmic interventions, such as fine-tuning or dataset curation, may be required to empower LLM-based decision making agents in complex settings.

cs.LG↗