Publications
Our teams aspire to make discoveries that impact everyone, and core to our approach is sharing our research and tools to fuel progress in the field.
Our teams aspire to make discoveries that impact everyone, and core to our approach is sharing our research and tools to fuel progress in the field.
Sort By
1 - 15 of 10795 publications
Preview abstract
AI coding assistants are rapidly becoming integral to modern software development. A key challenge in this space is the continual need to migrate and modernize codebases in response to evolving software ecosystems. Traditionally, such migrations have relied on rule-based systems and human intervention. With the advent of powerful large language models (LLMs), AI-driven agentic frameworks offer a promising alternative—but their effectiveness remains underexplored. In this paper, we introduce FreshBrew, a novel benchmark for evaluating AI-based agentic frameworks on project-level Java migrations. We benchmark several such frameworks, powered by state-of-the-art LLMs, and compare their performance against established rule-based tools. Our evaluation of AI agents on this benchmark of 228 repositories shows that the top-performing model, Gemini 2.5 Flash, can successfully migrate 56.5% of projects to JDK 17. Our empirical analysis reveals novel insights into the critical strengths and limitations of current agentic approaches, offering actionable insights into their real-world applicability. By releasing FreshBrew publicly upon acceptance, we aim to facilitate rigorous, reproducible evaluation and catalyze progress in AI-driven codebase modernization.
View details
Preview abstract
For many practical applications of quantum computing, the slowest and most costly steps involve coherently accessing classical data. We help address this challenge by applying mass production techniques, which can sometimes allow us to perform operations many times in parallel for a cost that is comparable to a single execution[1-3]. We combine existing mass-production results with modern approaches for loading classical data using ``quantum read-only memory.'' We show that quantum mass production techniques offer no benefit when we consider a cost model that focuses purely on the number of non-Clifford gates. However, analyzing the constant factors in a more nuanced cost model, we find that it may be possible to obtain a reduction in cost of an order or magnitude or more for a variety reasonably-sized fault-tolerant quantum algorithms. We present several applications of quantum mass-production techniques beyond naive parallelization, including a strategy for reducing the cost of serial calls to the same data loading step.
View details
PAIGE: Examining Student Learning Outcomes and Experiences with Personalized AI-Generated Podcasts
Tiffany Do
Usama Bin Shafqat
Elsie Ling
Νikhil Sarda
2025
Preview abstract
Generative AI is revolutionizing content creation and holds promise for real-time, personalized educational experiences. We investigated the effectiveness of converting textbook chapters into AI-generated podcasts and explored the impact of personalizing these podcasts
for individual learner profiles. We conducted a 3x3 user study with 180 college students in the United States, comparing traditional textbook reading with both generalized and personalized AI-generated podcasts across three textbook subjects. The personalized podcasts were tailored to students’ majors, interests, and learning styles. Our findings show that students found the AI-generated podcast format to be more enjoyable than textbooks and that personalized podcasts led to significantly improved learning outcomes, although this was subject-specific. These results highlight that AI-generated podcasts can offer an engaging and effective modality
transformation of textbook material, with personalization enhancing content relevance. We conclude with design recommendations for leveraging AI in education, informed by student feedback.
View details
Sufficient Context: A New Lens on Retrieval Augmented Generation Systems
Hailey Joren
Jianyi Zhang
Chun-Sung Ferng
Ankur Taly
International Conference on Learning Representations (ICLR) (2025)
Preview abstract
Augmenting LLMs with context leads to improved performance across many applications. Despite much research on Retrieval Augmented Generation (RAG) systems, an open question is whether errors arise because LLMs fail to utilize the context from retrieval or the context itself is insufficient to answer the query. To shed light on this, we develop a new notion of sufficient context, along with a method to classify instances that have enough information to answer the query. We then use sufficient context to analyze several models and datasets. By stratifying errors based on context sufficiency, we find that larger models with higher baseline performance (Gemini 1.5 Pro, GPT 4o, Claude 3.5) excel at answering queries when the context is sufficient, but often output incorrect answers instead of abstaining when the context is not. On the other hand, smaller models with lower baseline performance (Llama 3.1, Mistral 3, Gemma 2) hallucinate or abstain often, even with sufficient context. We further categorize cases when the context is useful, and improves accuracy, even though it does not fully answer the query and the model errs without the context. Building on our findings, we explore ways to reduce hallucinations in RAG systems, including a new selective generation method that leverages sufficient context information for guided abstention. Our method improves the fraction of correct answers among times where the model responds by 2--10% for Gemini, GPT, and Gemma.
View details
CASE: An Agentic AI Framework for Enhancing Scam Intelligence in Digital Payments
Jose Estevez
Shankey Poddar
Aviral Suri
Lorenzo Gatto
Zijun Kan
Diksha Bansal
Bill Cheung
2025
Preview abstract
The proliferation of digital payment platforms has transformed commerce, offering unmatched convenience and accessibility globally. However, this growth has also attracted malicious actors, leading to a corresponding increase in sophisticated social engineering scams. These scams are often initiated and orchestrated on multiple surfaces outside the payment platform, making user and transaction-based signals insufficient for a complete understanding of the scam's methodology and underlying patterns, without which it is very difficult to prevent it in a timely manner. This paper presents CASE (Conversational Agent for Scam Elucidation), a novel Agentic AI framework that addresses this problem by collecting and managing user scam feedback in a safe and scalable manner. A conversational agent is uniquely designed to proactively interview potential victims to elicit intelligence in the form of a detailed conversation. The conversation transcripts are then consumed by another AI system that extracts information and converts it into structured data for downstream usage in automated and manual enforcement mechanisms. Using Google's Gemini family of LLMs, we implemented this framework on Google Pay (GPay) India. By augmenting our existing features with this new intelligence, we have observed a 21% uplift in the volume of scam enforcements. The architecture and its robust evaluation framework are highly generalizable, offering a blueprint for building similar AI-driven systems to collect and manage scam intelligence in other sensitive domains.
View details
Preview abstract
This paper investigates the theoretical underpinnings of the widely successful
pretrain-then-adapt strategy for foundation models. We introduce a Bayesian
model selection criterion, termed the downstream free energy, which quantifies
the adaptability of a pretrained checkpoint by measuring, under the downstream
data distribution, the concentration of favorable solutions near the checkpoint.
However, minimizing this downstream free energy is infeasible without access to
downstream data. To address this, we show that under certain conditions, mini-
mizing the upstream free energy – which can be estimated using only upstream
data – can serve as a reliable proxy. We validate this theoretical insight through
preliminary experiments, showing that commonly used pretraining heuristics ef-
fectively lower upstream free energy, leading to better downstream performance.
View details
Spherical dimension
Bogdan Chornomaz
Shay Moran
Tom Waknine
2025
Preview abstract
We introduce and study the \emph{spherical dimension}, a natural topological relaxation of the VC dimension that unifies several results in learning theory where topology plays a key role in the proofs. The spherical dimension is defined by extending the set of realizable datasets (used to define the VC dimension) to the continuous space of realizable distributions. In this space, a shattered set of size d (in the VC sense) is completed into a continuous object, specifically a d-dimensional sphere of realizable distributions. The spherical dimension is then defined as the dimension of the largest sphere in this space. Thus, the spherical dimension is at least the VC dimension.
The spherical dimension serves as a common foundation for leveraging the Borsuk-Ulam theorem and related topological tools. We demonstrate the utility of the spherical dimension in diverse applications, including disambiguations of partial concept classes, reductions from classification to stochastic convex optimization, stability and replicability, and sample compression schemes. Perhaps surprisingly, we show that the open question posed by Alon, Hanneke, Holzman, and Moran (FOCS 2021) of whether there exist non-trivial disambiguations for halfspaces with margin is equivalent to the basic open question of whether the VC and spherical dimensions are finite together.
View details
Preview abstract
Large language models (LLMs), optimized through human feedback, have rapidly emerged as a leading paradigm for developing intelligent conversational assistants. However, despite their strong performance across many benchmarks, LLM-based agents might still lack conversational skills such as disambiguation -- when they are faced with ambiguity, they often overhedge or implicitly guess users' true intents rather than asking clarification questions. Under task-specific settings, high-quality conversation samples are often limited, constituting a bottleneck for LLMs' ability to learn optimal dialogue action policies. We propose Action-Based Contrastive Self-Training (ACT), a quasi-online preference optimization algorithm based on Direct Preference Optimization (DPO), that enables data-efficient dialogue policy learning in multi-turn conversation modeling. We demonstrate ACT's efficacy under data-efficient tuning scenarios, even when there is no action label available, using multiple real-world conversational tasks: tabular-grounded question-answering, machine reading comprehension, and AmbigSQL, a novel task for disambiguating information-seeking requests for complex SQL generation towards data analysis agents. Additionally, we propose evaluating LLMs' ability to function as conversational agents by examining whether they can implicitly recognize and reason about ambiguity in conversation. ACT demonstrates substantial conversation modeling improvements over standard tuning approaches like supervised fine-tuning and DPO.
View details
Data Quality Issues in Multilingual Speech Datasets: The Need for Sociolinguistic Awareness and Proactive Language Planning
Preview
Mingfei Lau
Allen Chen
Yeming Fang
Tingting Xu
Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics, Association for Computational Linguistics (ACL), Vienna, Austria (2025), 7466–7492
Private List Learnability vs. Online List Learnability
Hilla Schefler
Steve Hanneke
Iska Tsubari
Shay Moran
2025
Preview abstract
This work explores the connection between differential privacy (DP) and online learning in the context of PAC list learning. In this setting, a $k$-list learner outputs a list of $k$ potential predictions for an instance $x$ and incurs a loss if the true label of $x$ is not included in the list.
A basic result in the multiclass PAC framework with a finite number of labels states that private learnability is equivalent to online learnability [\citet*{AlonLMM19,BunLM20,JungKT20}].
Perhaps surprisingly, we show that this equivalence does not hold in the context of list learning.
Specifically, we prove that, unlike in the multiclass setting, a finite $k$-Littlestone dimension—a variant of the classical Littlestone dimension that characterizes online $k$-list learnability—is not a sufficient condition for DP $k$-list learnability.
However, similar to the multiclass case, we prove that it remains a necessary condition.
To demonstrate where the equivalence breaks down, we provide an example showing that the class of monotone functions with $k+1$ labels over $\mathbb{N}$ is online $k$-list learnable, but not DP $k$-list learnable.
This leads us to introduce a new combinatorial dimension, the \emph{$k$-monotone dimension}, which serves as a generalization of the threshold dimension.
Unlike the multiclass setting, where the Littlestone and threshold dimensions are finite together, for $k>1$, the $k$-Littlestone and $k$-monotone dimensions do not exhibit this relationship.
We prove that a finite $k$-monotone dimension is another necessary condition for DP $k$-list learnability, alongside finite $k$-Littlestone dimension.
Whether the finiteness of both dimensions implies private $k$-list learnability remains an open question.
View details
Mix&Slice
Marco Rosa
Encyclopedia of Cryptography, Security and Privacy, Springer Nature Switzerland (2025), pp. 1550-1555
Preview abstract
Mix&Slice is an encryption technique that enables efficient and robust access revocation on resources stored at external cloud providers. The technique makes use of a transformation that provides strong inter-dependency in the encrypted representation of a resource. To perform access revocation, it is then sufficient to re-encrypt a small portion of the resource to have guarantees that the resource (and any of its parts) will become unintelligible to those from whom access has been revoked.
View details
User-Centered Delivery of AI-Powered Health Care Technologies in Clinical Settings: Mixed Methods Case Study
Randall Brandt
Hien Brown
Christine Silva
JMIR Human Factors (2025)
Preview abstract
Background:
Providers spend a large percentage of their day using electronic health record (EHR) technology and frequently report frustration when EHR tasks are time-consuming and effortful. To solve these challenges, artificial intelligence (AI)–based enhancements to EHR technology are increasingly being deployed. However, AI-based implementations for EHR features often lack user-centered evaluation.
Objective:
This study evaluates, using a user-centered approach, the implementation of an AI-powered search and clinical discovery tool within an EHR system.
Methods:
We conducted a mixed methods study consisting of interviews, observations, and surveys for 5 months.
Results:
High adoption rates for the AI-based features (163/176, 93% users after 3 months) and significant increases across key metrics, including user satisfaction (U=49; P<.001) and perception of time saved (U=49; P<.001), demonstrated that the AI-based features were not only successfully integrated into various clinical workflows but also improved the user experience for clinicians.
Conclusions:
Our results underscore the feasibility and effectiveness of using a user-centered approach for the deployment of clinical AI tools. High adoption rates and positive user experiences were driven by our user-centered research program, which emphasized close collaboration with users, rapid incorporation of feedback, and tailored user training. This study program can be used as a starting framework for the design and integration of human-centered research methods for AI tool deployment in clinical settings.
View details
Record Number of Members Visit U.S. Congress to Talk Tech Policy
IEEE Spectrum (2025)
Preview abstract
This IEEE Spectrum article reflects on advocacy for U.S. technological leadership during my Congressional visit through IEEE-USA. Leading an expert group of other distinguished IEEE members, we urged lawmakers to support critical initiatives. Key priorities included sustained funding for federal research institutions like NIST, NASA, and the NSF, reauthorizing the SBIR/STTR programs vital for small business innovation, and passing the CREATE AI Act to democratize AI resources by establishing the National AI Research Resource (NAIRR).
We also emphasized strengthening the STEM talent pipeline through the CHIPS and Science Act and expanding high-skilled immigrant visas. We highlighted rapid AI advancements, such as autonomous vehicles, the surge in FDA-approved AI based medical devices, as underscoring the need for these strategic investments and policy actions. The article conveys a sense of urgency, calling for concrete congressional action to ensure the U.S. maintains its technological edge while also sharing my personal experiences.
View details
Preview abstract
Due to the size and complexity of modern large language models (LLMs), it has proven challenging to uncover the underlying mechanisms that models use to solve reasoning problems. For instance, is their reasoning for a specific problem localized to certain parts of the network? Do they break down the reasoning problem into modular components that are then executed as sequential steps as we go deeper in the model? To better understand the reasoning capability of LLMs, we study a minimal propositional logic problem that requires combining multiple facts to arrive at a solution. By studying this problem on Mistral and Gemma models, up to 27B parameters, we illuminate the core components the models use to solve such logic problems. From a mechanistic interpretability point of view, we use causal mediation analysis to uncover the pathways and components of the LLMs' reasoning processes. Then, we offer fine-grained insights into the functions of attention heads in different layers. We not only find a sparse circuit that computes the answer, but we decompose it into sub-circuits that have four distinct and modular uses. Finally, we reveal that three distinct models -- Mistral-7B, Gemma-2-9B and Gemma-2-27B -- contain analogous but not identical mechanisms.
View details
Preview abstract
Google has a long tradition of open-source software, which encompasses the field of operations research with OR-Tools. In development since 2008, it offers several solvers useful to many OR practitioners:
- PDLP, a revolutionary first-order linear solver that is reshaping the landscape of linear optimisation;
- CP-SAT, an award-winning constraint-programming solver;
- Glop, an accurate linear solver;
- Routing, a vehicle routing solver underpinning Google Maps Platform Route Optimization.
OR-Tools has long had its features accessible from other languages: the core algorithms are implemented in C++ for performance, but users can tap into them in Python, Java, C#, or Go.
It is recently available in Julia too, with a current focus on the linear and constraint solvers, either locally or remotely.
We provide a wrapper for our solvers that brings them to JuMP.jl through MathOptInterface.jl.
This tutorial will walk you through the features of OR-Tools and its solvers, then show examples of using OR-Tools from within Julia, either through JuMP or a lower-level interface.
We will also share our experience of C++-Julia interop.
View details
