Scientific research is being reshaped by AI systems that move beyond isolated assistance toward longer-horizon workflows spanning literature grounding, hypothesis generation, experimentation, validation, reporting, and revision. This shift marks a transition from task-level AI for science to workflow-level research automation. Yet current systems remain fragmented, differing in autonomy, domain scope, execution environment, validation mechanism, and human oversight, while still struggling with evidence preservation, reproducibility, weak-direction rejection, provenance tracking, cross-domain robustness, and accountable scientific closure. This survey examines these developments through AutoResearch, defined as the developmental spectrum of AI-powered scientific workflow automation. Within it, Vibe Research denotes the human-steered region of prompt-based assistance and human-verified execution, whereas emerging AI-led systems coordinate larger portions of the discovery loop without achieving robust autonomy. We analyze how research systems redistribute control, evidence, execution, validation, and accountability across workflows and organize the field around five workflow conditions: literature and research grounding; hypothesis formation and planning; experimentation and tool use; feedback, validation, and review; and reporting and knowledge communication. We further synthesize AI scientist systems, mixed-initiative co-research frameworks, benchmarks, domain deployments, and open-source infrastructures. Finally, we propose five evaluation dimensions–novelty, validity, impact, reliability, and provenance–and show that AutoResearch autonomy is domain-conditioned, being more credible in structured, executable, and rapidly verifiable settings but limited in embodied, delayed, heterogeneous, ethical, or institutionally accountable contexts.

Multimodal large language models (MLLMs) have demonstrated strong capabilities in visual understanding, yet they remain limited in complex, multi-step reasoning that requires deep searching and integrating visual evidence with external knowledge. In this work, we address this challenge by constructing high-quality, verified multi-hop vision-language training data for multimodal deep-search agents. We propose a Multi-hop Tool-Augmented Agent for Evidence-based QA Synthesis (MTA-Agent), which automatically selects tools and their parameters to retrieve and validate evidence from both visual and textual sources and generates structured multi-hop question-answer trajectories. Starting from diverse VQA seed datasets, our pipeline produces a large-scale training dataset, MTA-Vision-DeepSearch, containing 21K high-quality multi-hop examples. The data is filtered through a multi-stage verification process to ensure factual consistency and answer uniqueness. Using MTA-Vision-DeepSearch, a 32B open-source multimodal search agent achieves state-of-the-art performance, reaching an average of 54.63% across six challenging benchmarks, outperforming GPT-5 (51.86%), Gemini-2.5-Pro (50.98%), and Gemini-3-Pro (54.46%) under the same tool settings. We further show that training on our data improves both reasoning depth and tool-use behavior, increasing the average number of steps from 2.27 to 4.28, and leading to more systematic and persistent search strategies. Additionally, we demonstrate that training can be performed without real-time tool calls by replaying cached interactions, significantly reducing training cost. Importantly, we present MTA-Agent as a fully open recipe for multimodal deep search: we release the entire dataset, training trajectories, and implementation details to enable reproducibility and future research on open multimodal search agents.

Background Using digital tools to support healthy food choices and access may counteract the negative effects of living in under-resourced communities on diet quality. Objective We aimed to develop and evaluate the psychometric properties of a Digital Food and Nutrition Literacy (DFNL) survey and its association with diet quality for informing future tailored interventions to improve diet quality. Methods The DFNL survey was theory-informed by the Information-Motivation-Behavioral Skills Model and a multi-dimensional model for making healthy food choices online. In this cross-sectional study, 196 adults from a low-income, low food access community in Northeast Connecticut completed an online survey. Principal component analysis was used to identify a DFNL index and components for multivariable linear regression analyses with diet quality, an external variable theoretically expected to be associated with DFNL. Diet quality was assessed using the Short Healthy Eating Index (sHEI) and liking-based Diet Quality Index (DQI). Results The 10-item DFNL index was internally reliable and explained 53.8% of total variance through three interpretable components—online grocery shopping, online nutrition information, and online communication. Associations between DFNL and diet quality varied by DFNL component and diet quality measure; strongest associations were observed for online nutrition information and online grocery shopping components, particularly with adequacy scores of both diet quality measures. Conclusions These preliminary findings support the construct validity of the DFNL index. Assessing DFNL may help identify modifiable behavioral skills in community nutrition interventions that can be tailored to participants’ knowledge gaps and motivation to support diet quality in under-resourced populations. Summary The developed Digital Food and Nutrition Literacy survey showed initial evidence of construct validity. The three component DFNL structure was definable, showing significant associations with diet quality, particularly adequacy scores.

Geographic locations may influence social network characteristics and structures, and social networks may change one’s spatial access and mobility pattern. Such interdependence between social networks and spatial contexts is critical to consider when examining social determinants of health factors that act through them, especially in syndemics research focusing on interdependence and synergy among co-occurring diseases and social problems. In this chapter, we illustrate the dynamic process of such interdependencies and how they contribute to health inequity through differential exposure and differential vulnerability and propose a social-spatial network approach to examine health inequity in syndemics research through the lens of intersectionality.

Self-evolving agents requires adaptation after deployment, but existing approaches assume a usable learning loop, such as curated skills, successful trajectories, or verifier signals. Real open-world deployments may provide none of these, offering only a task prompt. In this work, we study open-world self-evolution, where an agent must build both its skills and its own verification signals from scratch, using open-world resources but no target-task supervision. We propose OpenSkill, a framework that bootstraps this loop: it acquires grounded knowledge and verification anchors from documentation, repositories, and the web, synthesizes them into transferable skills, and refines those skills against self-built virtual tasks grounded in the anchors rather than in target answers. The open world thus supplies both the knowledge to be learned and a supervision-independent practice environment, with target-task supervision reserved for final evaluation. Across three benchmarks and two target agents, OpenSkill attains the best automated pass rate while satisfying the no-supervision constraint. Analysis shows its skills transfer across models without model-specific adaptation, and its self-built verifier aligns with ground-truth outcomes despite never accessing them.

Autonomous agents that operate computers via Graphical User Interfaces (GUIs) often struggle with efficiency and reliability on complex, long-horizon tasks. While augmenting these agents with planners can improve task decomposition, they remain constrained by the inherent limitations of performing all actions through GUI manipulation, leading to brittleness and inefficiency. In this work, we introduce a more robust and flexible paradigm: enabling agents to use coding as an enhanced action. We present CoAct-1, a novel multi-agent system that synergistically combines GUI-based control with direct programmatic execution. CoAct-1 features an Orchestrator that dynamically delegates subtasks to either a conventional GUI Operator or a specialized Programmer agent, which can write and execute Python or Bash scripts. This hybrid approach allows the agent to bypass inefficient GUI action sequences for tasks like file management and data processing, while still utilizing visual interaction when necessary. We evaluate our system on the challenging OSWorld and WindowsAgentArena benchmark, where CoAct-1 achieves a new state-of-the-art success rate of 60.8% on OSWorld and 52.5% on WindowsAgentArena, significantly outperforming prior methods. Furthermore, our approach dramatically improves efficiency, reducing the average number of steps required to complete a task to just 10.15 on OSWorld, compared to 15 for leading GUI agents. Our results demonstrate that integrating coding as a core action provides a more powerful, efficient, and scalable path toward generalized computer automation.

Graphical user interface (GUI) agents autonomously complete tasks across platforms (\eg, Linux) by sequentially decomposing user instructions into action proposals that iteratively interact with visual elements in the evolving environment. However, two main challenges arise: i) planning (\ie, the action proposal sequence) under expansive action space, where selecting an appropriate plan is non-trivial, as many valid ones may exist; ii) accurately grounding actions in complex and high-resolution interfaces, \ie, precisely interacting with visual targets. This paper investigates the aforementioned challenges with our \textbf{G}UI \textbf{T}est-time Scaling \textbf{A}gent, namely \name. First, we conduct test-time scaling to select the most appropriate action proposal: at each step, multiple candidate proposals are sampled and evaluated and selected by a judge model. It trades off computation for better decision quality by concurrent sampling. Second, we propose a model that improves grounding of the selected action proposals to its corresponding visual elements. Our key insight is that reinforcement learning (RL) facilitates grounding through inherent objective alignments, rewarding successful clicks on interface elements. Experimentally, \name achieves state-of-the-art performance on both grounding and agent task execution benchmarks.

Multimodal embedding models have been crucial in enabling various downstream tasks such as semantic similarity, information retrieval, and clustering over different modalities. However, existing multimodal embeddings like VLM2Vec, E5-V, GME are predominantly focused on natural images, with limited support for other visual forms such as videos and visual documents. This restricts their applicability in real-world scenarios, including AI agents, multi-modal search and recommendation, and retrieval-augmented generation (RAG). To close this gap, we propose VLM2Vec-V2, a unified framework for learning embeddings across diverse visual forms. First, we introduce MMEB-V2, a comprehensive benchmark that extends MMEB with five new task types: visual document retrieval, video retrieval, temporal grounding, video classification and video question answering - spanning text, image, video, and visual document inputs. Next, we train VLM2Vec-V2, a general-purpose embedding model that supports text, image, video, and visual document inputs. Extensive experiments show that VLM2Vec-V2 achieves strong performance not only on the newly introduced video and document retrieval tasks, but also improves over prior baselines on the original image benchmarks. Through extensive evaluation, our study offers insights into the generalizability of various multimodal embedding models and highlights effective strategies for unified embedding learning, laying the groundwork for more scalable and adaptable representation learning in both research and real-world settings.

Current Video Large Language Models (Video LLMs) excel in question answering (QA) but largely operate as black boxes, providing textual answers without verifiable visual grounding. Existing explainability efforts rely on textual rationales or sparse bounding boxes, which struggle to capture complex video dynamics such as occlusions and non-rigid deformations. We propose Evidence-Backed Video Question Answering (E-VQA), a novel task requiring models to jointly output a semantic answer and precise spatio-temporal evidence: temporal segments and dense, tracked object segmentation masklets. To support this, we introduce ST-Evidence, the first human-verified benchmark for both discriminative and generative pixel-level grounding. Evaluations of state-of-the-art models reveal a critical decoupling between QA accuracy and true visual perception that scaling alone fails to bridge. To address this, we develop scalable, automated generation pipelines to create ST-Evidence-Instruct, a 160k-scale dataset bridging high-level reasoning with fine-grained grounding. Fine-tuning grounded Video LLMs on this data yields substantial gains over the corresponding size-matched UniPixel baselines (e.g., +27.2 t-mean and +13.8 J F on a 7B model), establishing a robust baseline for explainable, evidence-backed video understanding. Code and data are available at https://github.com/SalesforceAIResearch/EVQA.

Substantial variation in COVID-19 vaccination trends across U.S. communities raises key questions about the factors that shape vaccination uptake. We develop a dynamic simulation model that captures population-driven feedback processes, from social influence and uptake heterogeneity to fear of infection, responses to vaccine availability, and capacity constraints. To estimate behavioral parameters, we derive an ordinary differential equation (ODE)-informed Tobit regression framework and apply it to weekly vaccination data from all 50 US states and the District of Columbia during 2021-2022. The resulting estimates provide a robust empirical foundation for model calibration. The calibrated model reproduces observed regional vaccination trajectories and reveals three distinct phases of uptake: an early phase in which capacity and eligibility constraints bind on the underlying demand, a middle phase in which behavioral responses to changing COVID-19 incidence shape the fluctuations in uptake, with rising case numbers stimulating additional vaccination; and a late phase shaped by saturation dynamics. Finally, we document substantial regional heterogeneity in responsiveness to infection risk and the overall tendency to vaccinate when capacity permits.

The COVID-19 pandemic made many U.S. households susceptible to food insecurity and sparked a temporary expansion of federal food assistance. Recognizing that the effects of the pandemic on food insecurity were socially patterned and changed over time, this study aimed to identify different food insecurity trajectories from 2018 to 2022 and demographic/household factors associated with these trajectories in a low-income sample. We conducted a secondary analysis of data from a longitudinal annual survey (2018-2022) of 414 low-wage workers recruited in community settings in two U.S. cities: Raleigh, NC, and Minneapolis, MN. Annual survey measures included the 6-Item Food Security Module, self-reported demographics, and household economic factors including housing stability, employment, and receipt of Supplemental Nutrition Assistance Program (SNAP). Latent class analysis identified underlying classes (trajectories) of food security over time and assigned participants to a trajectory. Multinomial logistic regression models tested the association between demographic/household factors and trajectory membership. Food insecurity was high at baseline (72.7%). Latent class analysis yielded five trajectories from 2018 to 2022: (i) consistent high food security (23.9%), (ii) consistent moderate food security (28.3%), (iii) consistent very low food security (17.6%), (iv) improved food security (12.6%), and (v) temporarily improved food security (17.6%). Several demographic/household factors were associated with the likelihood of experiencing trajectories of less food security. Understanding trajectories of food insecurity during the COVID-19 pandemic can inform policy responses during future health or economic crises. Food security status is not a static condition; annual measurement is critical for promoting food security among low-income U.S. families.

Autonomous GUI agents face two fundamental challenges: early stopping, where agents prematurely declare success without verifiable evidence, and repetitive loops, where agents cycle through the same failing actions without recovery. We present VLAA-GUI, a modular GUI agentic framework built around three integrated components that guide the system on when to Stop, Recover, and Search. First, a mandatory Completeness Verifier enforces UI-observable success criteria and verification at every finish step – with an agent-level verifier that cross-examines completion claims with decision rules, rejecting those lacking direct visual evidence. Second, a mandatory Loop Breaker provides multi-tier filtering: switching interaction mode after repeated failures, forcing strategy changes after persistent screen-state recurrence, and binding reflection signals to strategy shifts. Third, an on-demand Search Agent searches online for unfamiliar workflows by directly querying a capable LLM with search ability, returning results as plain text. We additionally integrate a Coding Agent for code-intensive actions and a Grounding Agent for precise action grounding, both invoked on demand when required. We evaluate VLAA-GUI across five top-tier backbones, including Opus 4.5, 4.6 and Gemini 3.1 Pro, on two benchmarks with Linux and Windows tasks, achieving top performance on both (77.5

Web agents promise to automate complex browser tasks, but current methods remain brittle -- relying on step-by-step UI interactions and heavy LLM reasoning that break under dynamic layouts and long horizons. Humans, by contrast, exploit website-provided functionality through high-level operations like search, filter, and sort. We introduce WALT (Web Agents that Learn Tools), a framework that reverse-engineers latent website functionality into deterministic, callable tools. Rather than hypothesizing ad-hoc skills, WALT exposes robust implementations of automations already designed into websites, spanning discovery (search, filter, sort), communication (post, comment, upvote), and content management (create, edit, delete). Tools abstract away low-level execution: instead of reasoning about how to click and type, agents simply call search(query) or create(listing). This shifts the computational burden from fragile step-by-step reasoning to reliable tool invocation. On VisualWebArena and WebArena, WALT achieves significantly higher success rates with fewer steps and less LLM-dependent reasoning, establishing a robust and generalizable paradigm for browser automation.

Future motion representations, such as optical flow, offer immense value for control and generative tasks. However, forecasting generalizable spatially dense motion representations remains a key challenge, and learning such forecasting from noisy, real-world data remains relatively unexplored. We introduce FOFPred, a novel language-conditioned optical flow forecasting model featuring a unified Vision-Language Model (VLM) and Diffusion architecture. This unique combination enables strong multimodal reasoning with pixel-level generative fidelity for future motion prediction. Our model is trained on web-scale human activity data-a highly scalable but unstructured source. To extract meaningful signals from this noisy video-caption data, we employ crucial data preprocessing techniques and our unified architecture with strong image pretraining. The resulting trained model is then extended to tackle two distinct downstream tasks in control and generation. Evaluations across robotic manipulation and video generation under language-driven settings establish the cross-domain versatility of FOFPred, confirming the value of a unified VLM-Diffusion architecture and scalable learning from diverse web data for future optical flow prediction.

We introduce SCUBA, a benchmark designed to evaluate computer-use agents on customer relationship management (CRM) workflows within the Salesforce platform. SCUBA contains 300 task instances derived from real user interviews, spanning three primary personas—platform administrators, sales representatives, and service agents. The tasks test a range of enterprise-critical abilities, including Enterprise Software UI navigation, data manipulation, workflow automation, information retrieval, and troubleshooting. To ensure realism, SCUBA operates in Salesforce sandbox environments with support for parallel execution and fine-grained evaluation metrics to capture milestone progress. We benchmark a diverse set of agents under both zero-shot and demonstration-augmented settings. We observed huge performance gaps in different agent design paradigm and gaps between the open-source model and the closed-source model. In the zero-shot setting, open-source model powered computer-use agents that have strong performance on related benchmarks like OSWorld only have less than 5\% success rate on SCUBA, while methods built on closed-source models can still have up to 39\% percent task success rate. In the demonstration-augmented settings, task success rates can be improved to 50\% while simultaneously reducing time and costs by 13\% and 16\%, respectively. These findings highlight both the challenges of enterprise tasks automation and the promise of agentic solutions. By offering a realistic benchmark with interpretable evaluation, SCUBA aims to accelerate progress in building reliable computer-use agents for complex business software ecosystems.

The marketing literature has called for a better understanding of vulnerable consumer behavior as the purchasing power of these segments grows. However, current literature has not systematically examined the online review writing behavior of vulnerable consumer groups, creating a gap in our understanding of how frequently these consumers contribute their voices to influential online brand conversations. This work examines the representativeness of a restaurant's customer base using a custom dataset of online restaurant reviews combined with geospatial data on vulnerable consumers. We find that consumers in households with the elderly and/or disabled, racial/ethnic minorities, or suboptimal living conditions are more likely to write restaurant reviews, while consumers in households with low socioeconomic status are less likely to write reviews. However, only those consumers in suboptimal living conditions put more effort into their reviews. We suggest that these effects are driven by vulnerable consumers' perceived empowerment or lack thereof.

An LLM agent's capability depends not only on model weights but on its harness: prompts, tools, skills, and control flow. Self-improvement loops already edit harnesses, yet single-lineage search is path-dependent and local wins often regress other tasks. We introduce DarwinX, which treats self-evolution as selection over a population of harnesses with the model frozen: a preserve-and-extend contract admits only variants that extend coverage without regressing, an archive keeps alternative lineages for recombination, and failure-, teacher-, and self-derived evidence share one edit interface. Fitness comes from each benchmark's own verifier: no gold solutions, no hand-picked winners. Across four benchmarks that progressively separate the evolution signal from the test, one loop adds about 17 points on average: Terminal-Bench 2.1 rises +7.7 to 83.2