❌

Normal view

  • ✇MIT Technology Review
  • Deploying a hybrid approach to Web3 in the AI era MIT Technology Review Insights
    When the concept of “Web 3.0” first emerged about a decade ago the idea was clear: Create a more user-controlled internet that lets you do everything you can now, except without servers or intermediaries to manage the flow of information. Where Web2, which emerged in the early 2000s, relies on centralized systems to store data and supply compute, all owned—and monetized by—a handful of global conglomerates, Web3 turns that structure on its head. Instead, data and compute are decentralized thr
     

Deploying a hybrid approach to Web3 in the AI era

When the concept of “Web 3.0” first emerged about a decade ago the idea was clear: Create a more user-controlled internet that lets you do everything you can now, except without servers or intermediaries to manage the flow of information.

Where Web2, which emerged in the early 2000s, relies on centralized systems to store data and supply compute, all owned—and monetized by—a handful of global conglomerates, Web3 turns that structure on its head. Instead, data and compute are decentralized through technologies like blockchain and peer-to-peer networks.

What was once a futuristic concept is quickly becoming a more concrete reality, even at a time when Web2 still dominates. Six out of ten Fortune 500 companies are exploring blockchain-based solutions, most taking a hybrid approach that combines traditional Web2 business models and infrastructure with the decentralized technologies and principles of Web3.

Popular use cases include cloud services, supply chain management, and, most notably financial services. In fact, at one point, the daily volume of transactions processed on decentralized finance exchanges exceeded $10 billion.

Gaining a Web3 edge

Among the advantages of Web3 for the enterprise are greater ownership and control of sensitive data, says Erman Tjiputra, founder and CEO of the AIOZ Network, which is building infrastructure for Web3, powered by decentralized physical infrastructure networks (DePIN), blockchain-based systems that govern physical infrastructure assets.

More cost-effective compute is another benefit, as is enhanced security and privacy as the cyberattack landscape grows more hostile, he adds. And it could even help protect companies from outages caused by a single point of failure, which can lead to downtime, data loss, and revenue deficits.

But perhaps the most exciting opportunity, says Tjiputra, is the ability to build and scale AI reliably and affordably. By leveraging a people-powered internet infrastructure, companies can far more easily access—and contribute to—shared resource like bandwidth, storage, and processing power to run AI inference, train models, and store data. All while using familiar developer tooling and open, usage-based incentives.

“We’re in a compute crunch where requirements are insatiable, and Web3 creates this ability to benefit while contributing,” explains Tjiputra.

In 2025, AIOZ Network launched a distributed compute platform and marketplace where developers and enterprises can access and monetize AI assets, and run AI inference or training on AIOZ Network’s more than 300,000 contributing devices. The model allows companies to move away from opaque datasets and models and scale flexibly, without centralized lock in.

Overcoming Web3 deployment challenges

Despite the promise, it is still early days for Web3, and core systemic challenges are leaving senior leadership and developers hesitant about its applicability at scale.

One hurdle is a lack of interoperability. The current fragmentation of blockchain networks creates a segregated ecosystem that makes it challenging to transfer assets or data between platforms. This often complicates transactions and introduces new security risks due to the reliance on mechanisms such as cross-chain bridges. These are tools that allow asset transfers between platforms but which have been shown to be vulnerable to targeted attacks.

“We have countless blockchains running on different protocols and consensus models,” says Tjiputra. “These blockchains need to work with each other so applications can communicate regardless of which chain they are on. This makes interoperability fundamental.”

Regulatory uncertainty is also a challenge. Outdated legal frameworks can sit at odds with decentralized infrastructures, especially when it comes to compliance with data protection and anti-money laundering regulations.

“Enterprises care about verifiability and compliance as much as innovation, so we need frameworks where on-chain transparency strengthens accountability instead of adding friction,” Tjiputra says.

And this is compounded by user experience (UX) challenges, says Tjiputra. “The biggest setback in Web3 today is UX,” he says. “For example, in Web2, if I forget my bank username or password, I can still contact the bank, log in and access my assets. The trade-off in Web3 is that, should that key be compromised or lost, we lose access to those assets. So, key recovery is a real problem.”

Building a bridge to Web3

Although such systemic challenges won’t be solved overnight, by leveraging DePIN networks, enterprises can bridge the gap between Web2 and Web3, without making a wholesale switch. This can minimize risk while harnessing much of the potential.

AIOZ Network’s own ecosystem includes capacity for media streaming, AI compute, and distributed storage that can be plugged into an existing Web2 tech stack. “You don’t need to go full Web3,” says Tjiputra. “You can start by plugging distributed storage into your workflow, test it, measure it, and see the benefits firsthand.”

The AIOZ Storage solution, for example, offers scalable distributed object storage by leveraging the global network of contributor devices on AIOZ DePIN. It is also compatible with existing storage systems or commonly used web application programming interfaces (APIs).

“Say we have a programmer or developer who uses Amazon S3 Storage or REST APIs, then all they need to do is just repoint the endpoints,” explains Tjiputra. “That’s it. It’s the same tools, it’s really simple. Even with media, with a single one-stop shop, developers can do transcoding and streaming with a simple REST API.”

Built on Cosmos, a network of hundreds of different blockchains that can communicate with each other, and a standardized framework enabled by Ethereum Virtual Machine (EVM), AIOZ Network has also prioritized interoperability. “Applications shouldn’t care which chain they’re on. Developers should target APIs without worrying about consensus mechanisms. That’s why we built on Cosmos and EVM—interoperability first.”

This hybrid model, which allows enterprises to use both Web2 and Web3 advantages in tandem, underpins what Tjiputra sees as the longer-term ambition for the much-hyped next iteration of the internet.

“Our vision is a truly peer-to-peer foundation for a people-powered internet, one that minimizes single points of failure through multi-region, multi-operator design,” says Tjiputra. “By distributing compute and storage across contributors, we gain both cost efficiency and end-to-end security by default.

“Ideally, we want to evolve the internet toward a more people-powered model, but we’re not there yet. We’re still at the starting point and growing.”

Indeed, Web3 isn’t quite snapping at the heels of the world’s Web2 giants, but its commercial advantages in an era of AI have become much harder to ignore. And with DePIN bridging the gap, enterprises and developers can step into that potential while keeping one foot on surer ground.

To learn more from AIOZ Network, you can read the AIOZ Network Vision Paper.

This content was produced by Insights, the custom content arm of MIT Technology Review. It was not written by MIT Technology Review’s editorial staff.

This content was researched, designed, and written by human writers, editors, analysts, and illustrators. This includes the writing of surveys and collection of data for surveys. AI tools that may have been used were limited to secondary production processes that passed thorough human review.

Adaptive therapy for perioperative non-small cell lung cancer: strategies guided by dynamic minimal residual disease adjustment

Transl Oncol. 2026 Jan 6;64:102660. doi: 10.1016/j.tranon.2025.102660. Online ahead of print.

ABSTRACT

Lung cancer remains the leading cause of cancer incidence and mortality worldwide, with non-small cell lung cancer (NSCLC) accounting for about 85% of cases. The low rate of early diagnosis and the high rate of occult metastases limit the survival benefits of conventional treatments. The current TNM staging system fails to fully reflect tumor heterogeneity or the dynamic molecular evolution of the disease, thus affecting the prediction of recurrence and the prognostic stratification. Some recent advances in minimal residual disease (MRD) detection, such as ultra-sensitive liquid biopsy technologies, have largely overcome the limitations of traditional imaging and offered a transformative approach for continuous, precision-based management of lung cancer. This review systematically summarized the technological evolution of MRD detection and highlighted its clinical significance in guiding adaptive therapy for NSCLC, including treatment escalation, de-escalation, and the emerging concept of precision-guided drug holidays. Moreover, the authors comprehensively discussed the "Four-Dimensional TNMB Staging System," which incorporates continuous molecular monitoring to address the static limitations of conventional staging and enhance the accuracy of prognostic stratification. Although ongoing challenges, such as the lack of standardized interpretation criteria and limited detection sensitivity, the combinations with the third-generation liquid biopsy platforms, multi-omics analyses, and multi-center prospective validation studies are expected to advance the clinical implementation of MRD-guided strategies. The paradigm change will enable the transition of NSCLC management from conventional standardized models to a precision-guided, closed-loop system of "monitoring-intervention-remonitoring," establishing a solid theoretical and practical foundation for comprehensive, molecularly driven management strategies.

PMID:41496417 | DOI:10.1016/j.tranon.2025.102660

The Path Ahead for Agentic AI: Challenges and Opportunities

arXiv:2601.02749v1 Announce Type: new Abstract: The evolution of Large Language Models (LLMs) from passive text generators to autonomous, goal-driven systems represents a fundamental shift in artificial intelligence. This chapter examines the emergence of agentic AI systems that integrate planning, memory, tool use, and iterative reasoning to operate autonomously in complex environments. We trace the architectural progression from statistical models to transformer-based systems, identifying capabilities that enable agentic behavior: long-range reasoning, contextual awareness, and adaptive decision-making. The chapter provides three contributions: (1) a synthesis of how LLM capabilities extend toward agency through reasoning-action-reflection loops; (2) an integrative framework describing core components perception, memory, planning, and tool execution that bridge LLMs with autonomous behavior; (3) a critical assessment of applications and persistent challenges in safety, alignment, reliability, and sustainability. Unlike existing surveys, we focus on the architectural transition from language understanding to autonomous action, emphasizing the technical gaps that must be resolved before deployment. We identify critical research priorities, including verifiable planning, scalable multi-agent coordination, persistent memory architectures, and governance frameworks. Responsible advancement requires simultaneous progress in technical robustness, interpretability, and ethical safeguards to realize potential while mitigating risks of misalignment and unintended consequences.

AI-exposed jobs deteriorated before ChatGPT

arXiv:2601.02554v1 Announce Type: cross Abstract: Public debate links worsening job prospects for AI-exposed occupations to the release of ChatGPT in late 2022. Using monthly U.S. unemployment insurance records, we measure occupation- and location-specific unemployment risk and find that risk rose in AI-exposed occupations beginning in early 2022, months before ChatGPT. Analyzing millions of LinkedIn profiles, we show that graduate cohorts from 2021 onward entered AI-exposed jobs at lower rates than earlier cohorts, with gaps opening before late 2022. Finally, from millions of university syllabi, we find that graduates taking more AI-exposed curricula had higher first-job pay and shorter job searches after ChatGPT. Together, these results point to forces pre-dating generative AI and to the ongoing value of LLM-relevant education.

PatentMind: A Multi-Aspect Reasoning Graph for Patent Similarity Evaluation

arXiv:2505.19347v3 Announce Type: replace Abstract: Patent similarity evaluation plays a critical role in intellectual property analysis. However, existing methods often overlook the intricate structure of patent documents, which integrate technical specifications, legal boundaries, and application contexts. We introduce PatentMind, a novel framework for patent similarity assessment based on a Multi-Aspect Reasoning Graph (MARG). PatentMind decomposes patents into their three dimensions of technical features, application domains, and claim scopes, then dimension-specific similarity scores are calculated over the MARG. These scores are dynamically weighted through a context-aware reasoning process, which integrates contextual signals to emulate expert-level judgment. To support evaluation, we construct a human-annotated benchmark PatentSimBench, comprising 500 patent pairs. Experimental results demonstrate that the PatentMind-generated scores show a strong correlation ($r=0.938$) with expert annotations, significantly outperforming embedding-based models, patent-specific models, and advanced prompt engineering methods. Beyond computational linguistics, our framework provides a structured and semantically grounded foundation for real-world decision-making, particularly for tasks such as infringement risk assessment, underscoring its broader impact on both patent analytics and evaluation.

Organoids in translation: a bench-to-bedside framework for pancreatic cancer precision medicine

J Transl Med. 2026 Jan 6. doi: 10.1186/s12967-025-07596-8. Online ahead of print.

ABSTRACT

INTRODUCTION: Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal malignancies with a 5-year survival rate of < 13%. Standard treatments such as FOLFIRINOX or gemcitabine/nab-paclitaxel yield modest response rates, underscoring the urgent need for precision oncology approaches. Patient-derived organoids (PDOs) preserve the genomic, phenotypic, and histopathological features of the source tumor and offer a promising platform for drug screening, biomarker development, and personalized therapy. However, a systematic evaluation of their translational capacities is lacking.

METHODS: A systematic review was conducted according to the PRISMA 2020 guidelines (PROSPERO registration pending) using PubMed, EMBASE, and Cochrane CENTRAL (December 10, 2024) to identify English-language PDAC PDO studies that incorporated therapeutic testing. Ninety-five studies met the inclusion criteria. Data extraction captured >75 variables per study, including spanning culture methodology, therapeutic profiling, biomarker integration, and clinical correlation. A 13-domain weighted Translatability Scoring Framework adapted from Wehling et al. assessed predictive validity, biomarker strength, pharmacogenetics, and clinical trial alignment. Scores ranged from 0 to 5 and were categorized as good (>4.0), moderate (3.0-4.0), or low (<3.0) translational potential.

RESULTS: Of the 95 studies, 70.5% have been published since 2021, reflecting the rapid growth in this field. The mean PDO generation success rate was 89.7%, with the primary tumor tissue being the predominant source (48.4%). Only 24.8% were directly linked to clinical trials and 5.3% incorporated multi-omic profiling. The median translatability score was 3.13 (range, 1.72-4.59): 45.3% of the studies had low translatability, 50.5% moderate, and only 4.2% had good translational potential. High-scoring studies consistently combine multi-omic biomarker platforms, in vivo validation, clinical outcome correlation, and prospective trial integration. Conversely, the weakest domains were pharmacogenetics, endpoint strategies, and biomarker validation, limiting their overall clinical relevance.

CONCLUSIONS: PDOs have demonstrated strong feasibility and in vitro clinical correlation in PDAC; however, their clinical translation remains constrained by limited multi-omic integration, absence of pharmacogenomic modeling, and sparse clinical trial embedding. Standardization of protocols, adoption of harmonized and clinically relevant endpoints, and systematic incorporation of biomarker-driven co-clinical trial frameworks are urgently needed to transition PDOs from promising experimental surrogates to validating precision oncology tools capable of informing therapeutic decision-making in PDAC.

PMID:41495743 | DOI:10.1186/s12967-025-07596-8

Establishment and Optimization of a Patient-Reported Outcome–Based Electronic-Diary for Symptoms Evaluation in Patients With Gastroesophageal Reflux Disorder: Prospective Cohort Study

Background: Gastroesophageal reflux disease (GERD) symptoms significantly affect patients’ quality of life. Patient-reported outcome (PRO) instruments for symptoms measurement in GERD patients is advocated by regulatory authority. Current tools for GERD symptoms evaluation are limited and the results can be biased by the recall bias. To better characterize the GERD symptoms, an e-diary was developed for daily GERD symptom monitoring. Objective: To build up and optimize a PRO-based e-diary, and to investigate the effect of symptom frequency on adherence. Methods: The GERD e-diary evaluated 8 daytime (acid regurgitation, cough, heartburn, sour taste in the mouth, hiccups, hoarseness, dysphagia, and chest pain) and 2 nighttime symptoms (acid regurgitation and cough) for consecutive 8 weeks. The adherence of e-diary, defined as daily completing rate of e-diary, was evaluated and optimized from First Stage to Third Stage with no reminder implemented in First Stage, sending reminding SMS (Short Message Service) text messaging upon detecting missing data in Second Stage (no reminder during the first 3 to 5 days after enrollment), and immediate installation of reminding system at enrollment in Third Stage. GERD symptom frequency was obtained by summation of the symptomatic days in each week. A multiple regression analysis was performed to examine the effects of system optimization and GERD symptom frequency on patient adherence, while controlling for potential confounding variables. Results: 138 GERD patients (M/F=70/68; age: mean 52.9, SD 12.3 years) were recruited. At First Stage, the adherence was 47.2%, 40% and 57.6% for nighttime, daytime and overall symptom. System optimization significantly improved adherence with increased adherence of nighttime symptoms by 12.5% (P=.005) and 10.9% (P=.01), daytime symptom by 21.7% (P

STAT+: FDA announces sweeping changes to oversight of wearables, AI-enabled devices

7 January 2026 at 04:49

LAS VEGAS — The Food and Drug Administration announced Tuesday that it will ease regulation of digital health products, following through on the Trump administration’s promises to deregulate artificial intelligence and promote its widespread use.

FDA Commissioner Marty Makary indicated that one of the agency’s priorities is fostering an environment that’s good for investors, and that FDA regulation needs to move “at Silicon Valley speed.” He announced the changes during an address to conference attendees at the Consumer Electronics Show.

The agency will soften its approach to the regulation of clinical decision support software, which include AI-enabled products that help doctors navigate diagnoses and treatment options. The agency previously considered products that delivered a single recommendation as FDA-regulated medical devices. Now, those products can enter the market without FDA review as long as they fulfill the agency’s other criteria for escaping regulation. 

Continue to STAT+ to read the full story…

© ANDREW CABALLERO-REYNOLDS/AFP via Getty Images

California lawmaker proposes a four-year ban on AI chatbots in kids’ toys

7 January 2026 at 04:22
“Our children cannot be used as lab rats for Big Tech to experiment on,” Senator Steve Padilla said. He just introduced a bill to ban AI chatbots in toys until safety regulations are developed.  

The role of PCMT1 in prognosis tumor immune microenvironment and therapeutic responses across cancers

Discov Oncol. 2026 Jan 5. doi: 10.1007/s12672-025-04366-2. Online ahead of print.

ABSTRACT

BACKGROUND: Emerging evidence highlights the overexpression of Protein-L-isoaspartate (D-aspartate) O-methyltransferase (PCMT1) in multiple malignancies. However, its pan-cancer prognostic significance, tumor immune microenvironment (TIME) interactions, and therapeutic implications remain underexplored.

METHODS: Multi-omics data were integrated from UCSC Xena, GTEx, UALCAN, and published cohorts. PCMT1 expression patterns were systematically analyzed across 33 cancer types. Associations between PCMT1 and clinical outcomes, immune infiltration, immune checkpoint genes (ICGs), tumor mutation burden (TMB), microsatellite instability (MSI), and drug sensitivity were evaluated using bioinformatics pipelines.

RESULTS: Our pan-cancer analysis revealed differential expression patterns of PCMT1 across various malignancies, with significant upregulation in 20 cancer types and downregulation in 3 cancer types. Notably, PCMT1 overexpression was predominantly observed in epithelial-origin tumors, such as ACC (adrenocortical carcinoma), BRCA (breast invasive carcinoma), COAD (colon adenocarcinoma), and LUAD (lung adenocarcinoma). Survival analysis demonstrated that elevated PCMT1 expression was significantly correlated with unfavorable prognosis in multiple epithelial tumors, particularly in BRCA, esophageal carcinoma (ESCA), head and neck squamous cell carcinoma (HNSC), liver hepatocellular carcinoma (LIHC), and mesothelioma (MESO). Furthermore, comprehensive analysis identified significant associations between PCMT1 expression and various tumor microenvironment features, including immune scores, six distinct immune cell types, four immunosuppressive cell populations, cancer-associated fibroblasts (CAFs)-related markers, and immunosuppressive factors. PCMT1 expression also showed significant correlations with tumor mutation burden (TMB), microsatellite instability (MSI), DNA stemness score (DNAss), and RNA stemness score (RNAss). Particularly noteworthy was the strong positive correlation between PCMT1 expression and CAFs infiltration, along with their associated factors. These findings were further validated in independent immunotherapy cohorts, where PCMT1 consistently demonstrated immunosuppressive characteristics.

CONCLUSION: Multi-omics analysis suggests that PCMT1 may serve as a potential prognostic biomarker and a novel immunotherapy target for pan-cancer.

PMID:41491065 | DOI:10.1007/s12672-025-04366-2

PRIME: an interpretable artificial intelligence model based on liquid biopsy improves prediction of progression risk in non-small cell lung cancer

Mil Med Res. 2026 Jan 6;12(1):94. doi: 10.1186/s40779-025-00679-z.

ABSTRACT

BACKGROUND: Despite the predictive impact of circulating tumor DNA (ctDNA) minimal residual disease (MRD), accurate prediction of failure risk after curative-intent treatments for early-stage or localized non-small cell lung cancer (NSCLC) patients to guide personalized therapy remains challenging. This study aimed to develop and validate an interpretable artificial intelligence-assisted model using global data resources.

METHODS: Liquid biopsy data, blood-based genomic alterations, clinicopathological features, and survival outcomes of stage I-III NSCLC patients who underwent surgery or definitive chemoradiotherapy were collected from 6 cohorts. PRIME (Progression Risk prediction by Interpretable Machine learning on ctDNA-MRD, Mutations, and clinical-therapeutic features) was trained by 6 machine learning algorithms across 4 cohorts and validated in 2 independent cohorts. Model performance was evaluated by the area under the curve (AUC) and interpreted by SHapley Additive exPlanations (SHAP). Whole-exome sequencing (WES) or whole-genome sequencing (WGS) of tumor tissue from 430 stage II-III NSCLC patients and RNA-sequencing (RNA-seq) data from 1149 subjects, sourced from The Cancer Genome Atlas, were used to validate the prognostic effect of mutations identified in peripheral blood and investigate the underlying mechanisms.

RESULTS: A global dataset encompassing 781 blood samples from 493 patients was analyzed. Clinical stage, pre-treatment ctDNA, post-treatment MRD, blood-based Kelch-like ECH-associated protein 1 (KEAP1), serine/threonine kinase 11 (STK11), and cyclin-dependent kinase inhibitor 2A (CDKN2A) mutations, and treatment modality were significantly associated with the risk of disease progression and were thereby included in the model training. WES/WGS and RNA-seq confirmed the poor prognostic effect of KEAP1, STK11, and CDKN2A mutations, which were characterized by the suppressive tumor microenvironment and attenuated humoral immunity. The neural network (NN) model exhibited optimal prediction of treatment failure risk in the training (AUC = 0.85, 95% CI 0.81-0.89) and validation sets (AUC = 0.82, 95% CI 0.74-0.89). SHAP analysis indicated that MRD (+0.306), treatment modality (+0.128), and pre-treatment ctDNA (+0.043) ranked in the top 3 contributions. NN-PRIME outperformed single liquid biopsy biomarkers and clinical-therapeutic signatures, and demonstrated consistent robustness across different clinical scenarios. High-risk patients identified by NN-PRIME had poorer prognoses but derived significant benefits from adjuvant therapy after surgery.

CONCLUSIONS: As an interpretable model integrating readily-accessible and crucial clinical-genomic predictors, PRIME achieves enhanced performance, allowing for early outcome prediction, refined risk stratification, and personalized clinical decision-making.

PMID:41491583 | PMC:PMC12771999 | DOI:10.1186/s40779-025-00679-z

Digital Twin AI: Opportunities and Challenges from Large Language Models to World Models

arXiv:2601.01321v1 Announce Type: new Abstract: Digital twins, as precise digital representations of physical systems, have evolved from passive simulation tools into intelligent and autonomous entities through the integration of artificial intelligence technologies. This paper presents a unified four-stage framework that systematically characterizes AI integration across the digital twin lifecycle, spanning modeling, mirroring, intervention, and autonomous management. By synthesizing existing technologies and practices, we distill a unified four-stage framework that systematically characterizes how AI methodologies are embedded across the digital twin lifecycle: (1) modeling the physical twin through physics-based and physics-informed AI approaches, (2) mirroring the physical system into a digital twin with real-time synchronization, (3) intervening in the physical twin through predictive modeling, anomaly detection, and optimization strategies, and (4) achieving autonomous management through large language models, foundation models, and intelligent agents. We analyze the synergy between physics-based modeling and data-driven learning, highlighting the shift from traditional numerical solvers to physics-informed and foundation models for physical systems. Furthermore, we examine how generative AI technologies, including large language models and generative world models, transform digital twins into proactive and self-improving cognitive systems capable of reasoning, communication, and creative scenario generation. Through a cross-domain review spanning eleven application domains, including healthcare, aerospace, smart manufacturing, robotics, and smart cities, we identify common challenges related to scalability, explainability, and trustworthiness, and outline directions for responsible AI-driven digital twin systems.

Beyond Gemini-3-Pro: Revisiting LLM Routing and Aggregation at Scale

arXiv:2601.01330v1 Announce Type: new Abstract: Large Language Models (LLMs) have rapidly advanced, with Gemini-3-Pro setting a new performance milestone. In this work, we explore collective intelligence as an alternative to monolithic scaling, and demonstrate that open-source LLMs' collaboration can surpass Gemini-3-Pro. We first revisit LLM routing and aggregation at scale and identify three key bottlenecks: (1) current train-free routers are limited by a query-based paradigm focusing solely on textual similarity; (2) recent aggregation methods remain largely static, failing to select appropriate aggregators for different tasks;(3) the complementarity of routing and aggregation remains underutilized. To address these problems, we introduce JiSi, a novel framework designed to release the full potential of LLMs' collaboration through three innovations: (1) Query-Response Mixed Routing capturing both semantic information and problem difficulty; (2) Support-Set-based Aggregator Selection jointly evaluating the aggregation and domain capacity of aggregators; (3) Adaptive Routing-Aggregation Switch dynamically leveraging the advantages of routing and aggregation. Comprehensive experiments on nine benchmarks demonstrate that JiSi can surpass Gemini-3-Pro with only 47% costs by orchestrating ten open-source LLMs, while outperforming mainstream baselines. It suggests that collective intelligence represents a novel path towards Artificial General Intelligence (AGI).

Yuan3.0 Flash: An Open Multimodal Large Language Model for Enterprise Applications

arXiv:2601.01718v1 Announce Type: new Abstract: We introduce Yuan3.0 Flash, an open-source Mixture-of-Experts (MoE) MultiModal Large Language Model featuring 3.7B activated parameters and 40B total parameters, specifically designed to enhance performance on enterprise-oriented tasks while maintaining competitive capabilities on general-purpose tasks. To address the overthinking phenomenon commonly observed in Large Reasoning Models (LRMs), we propose Reflection-aware Adaptive Policy Optimization (RAPO), a novel RL training algorithm that effectively regulates overthinking behaviors. In enterprise-oriented tasks such as retrieval-augmented generation (RAG), complex table understanding, and summarization, Yuan3.0 Flash consistently achieves superior performance. Moreover, it also demonstrates strong reasoning capabilities in domains such as mathematics, science, etc., attaining accuracy comparable to frontier model while requiring only approximately 1/4 to 1/2 of the average tokens. Yuan3.0 Flash has been fully open-sourced to facilitate further research and real-world deployment: https://github.com/Yuan-lab-LLM/Yuan3.0.

The Qualitative Laboratory: Theory Prototyping and Hypothesis Generation with Large Language Models

arXiv:2601.00797v1 Announce Type: cross Abstract: A central challenge in social science is to generate rich qualitative hypotheses about how diverse social groups might interpret new information. This article introduces and illustrates a novel methodological approach for this purpose: sociological persona simulation using Large Language Models (LLMs), which we frame as a "qualitative laboratory". We argue that for this specific task, persona simulation offers a distinct advantage over established methods. By generating naturalistic discourse, it overcomes the lack of discursive depth common in vignette surveys, and by operationalizing complex worldviews through natural language, it bypasses the formalization bottleneck of rule-based agent-based models (ABMs). To demonstrate this potential, we present a protocol where personas derived from a sociological theory of climate reception react to policy messages. The simulation produced nuanced and counter-intuitive hypotheses - such as a conservative persona's rejection of a national security frame - that challenge theoretical assumptions. We conclude that this method, used as part of a "simulation then validation" workflow, represents a superior tool for generating deeply textured hypotheses for subsequent empirical testing.

MACA: A Framework for Distilling Trustworthy LLMs into Efficient Retrievers

arXiv:2601.00926v1 Announce Type: cross Abstract: Modern enterprise retrieval systems must handle short, underspecified queries such as ``foreign transaction fee refund'' and ``recent check status''. In these cases, semantic nuance and metadata matter but per-query large language model (LLM) re-ranking and manual labeling are costly. We present Metadata-Aware Cross-Model Alignment (MACA), which distills a calibrated metadata aware LLM re-ranker into a compact student retriever, avoiding online LLM calls. A metadata-aware prompt verifies the teacher's trustworthiness by checking consistency under permutations and robustness to paraphrases, then supplies listwise scores, hard negatives, and calibrated relevance margins. The student trains with MACA's MetaFusion objective, which combines a metadata conditioned ranking loss with a cross model margin loss so it learns to push the correct answer above semantically similar candidates with mismatched topic, sub-topic, or entity. On a proprietary consumer banking FAQ corpus and BankFAQs, the MACA teacher surpasses a MAFA baseline at Accuracy@1 by five points on the proprietary set and three points on BankFAQs. MACA students substantially outperform pretrained encoders; e.g., on the proprietary corpus MiniLM Accuracy@1 improves from 0.23 to 0.48, while keeping inference free of LLM calls and supporting retrieval-augmented generation.

Correctness isnt Efficiency: Runtime Memory Divergence in LLM-Generated Code

arXiv:2601.01215v1 Announce Type: cross Abstract: Large language models (LLMs) can generate programs that pass unit tests, but passing tests does not guarantee reliable runtime behavior. We find that different correct solutions to the same task can show very different memory and performance patterns, which can lead to hidden operational risks. We present a framework to measure execution-time memory stability across multiple correct generations. At the solution level, we introduce Dynamic Mean Pairwise Distance (DMPD), which uses Dynamic Time Warping to compare the shapes of memory-usage traces after converting them into Monotonic Peak Profiles (MPPs) to reduce transient noise. Aggregating DMPD across tasks yields a model-level Model Instability Score (MIS). Experiments on BigOBench and CodeContests show substantial runtime divergence among correct solutions. Instability often increases with higher sampling temperature even when pass@1 improves. We also observe correlations between our stability measures and software engineering indicators such as cognitive and cyclomatic complexity, suggesting links between operational behavior and maintainability. Our results support stability-aware selection among passing candidates in CI/CD to reduce operational risk without sacrificing correctness. Artifacts are available.

OpenNovelty: An LLM-powered Agentic System for Verifiable Scholarly Novelty Assessment

arXiv:2601.01576v1 Announce Type: cross Abstract: Evaluating novelty is critical yet challenging in peer review, as reviewers must assess submissions against a vast, rapidly evolving literature. This report presents OpenNovelty, an LLM-powered agentic system for transparent, evidence-based novelty analysis. The system operates through four phases: (1) extracting the core task and contribution claims to generate retrieval queries; (2) retrieving relevant prior work based on extracted queries via semantic search engine; (3) constructing a hierarchical taxonomy of core-task-related work and performing contribution-level full-text comparisons against each contribution; and (4) synthesizing all analyses into a structured novelty report with explicit citations and evidence snippets. Unlike naive LLM-based approaches, \textsc{OpenNovelty} grounds all assessments in retrieved real papers, ensuring verifiable judgments. We deploy our system on 500+ ICLR 2026 submissions with all reports publicly available on our website, and preliminary analysis suggests it can identify relevant prior work, including closely related papers that authors may overlook. OpenNovelty aims to empower the research community with a scalable tool that promotes fair, consistent, and evidence-backed peer review.
❌