arXiv:2512.05356v1 Announce Type: new Abstract: Self-improvement is a goal currently exciting the field of AI, but is fraught with danger, and may take time to fully achieve. We advocate that a more achievable and better goal for humanity is to maximize co-improvement: collaboration between human researchers and AIs to achieve co-superintelligence. That is, specifically targeting improving AI systems' ability to work with human researchers to conduct AI research together, from ideation to experimentation, in order to both accelerate AI research and to generally endow both AIs and humans with safer superintelligence through their symbiosis. Focusing on including human research improvement in the loop will both get us there faster, and more safely.

arXiv:2512.05365v1 Announce Type: new Abstract: Healthcare AI systems have historically faced challenges in merging contextual reasoning, long-term state management, and human-verifiable workflows into a cohesive framework. This paper introduces a completely innovative architecture and concept: combining the Model Context Protocol (MCP) with a specific clinical application, known as MCP-AI. This integration allows intelligent agents to reason over extended periods, collaborate securely, and adhere to authentic clinical logic, representing a significant shift away from traditional Clinical Decision Support Systems (CDSS) and prompt-based Large Language Models (LLMs). As healthcare systems become more complex, the need for autonomous, context-aware clinical reasoning frameworks has become urgent. We present MCP-AI, a novel architecture for explainable medical decision-making built upon the Model Context Protocol (MCP) a modular, executable specification for orchestrating generative and descriptive AI agents in real-time workflows. Each MCP file captures clinical objectives, patient context, reasoning state, and task logic, forming a reusable and auditable memory object. Unlike conventional CDSS or stateless prompt-based AI systems, MCP-AI supports adaptive, longitudinal, and collaborative reasoning across care settings. MCP-AI is validated through two use cases: (1) diagnostic modeling of Fragile X Syndrome with comorbid depression, and (2) remote coordination for Type 2 Diabetes and hypertension. In either scenario, the protocol facilitates physician-in-the-loop validation, streamlines clinical processes, and guarantees secure transitions of AI responsibilities between healthcare providers. The system connects with HL7/FHIR interfaces and adheres to regulatory standards, such as HIPAA and FDA SaMD guidelines. MCP-AI provides a scalable basis for interpretable, composable, and safety-oriented AI within upcoming clinical environments.

arXiv:2512.05449v1 Announce Type: new Abstract: Large language models display a peculiar form of inconsistency: they "know" the correct answer but fail to act on it. In human philosophy, this tension between global judgment and local impulse is called akrasia, or weakness of will. We propose akrasia as a foundational concept for analyzing inconsistency and goal drift in agentic AI systems. To operationalize it, we introduce a preliminary version of the Akrasia Benchmark, currently a structured set of prompting conditions (Baseline [B], Synonym [S], Temporal [T], and Temptation [X]) that measures when a model's local response contradicts its own prior commitments. The benchmark enables quantitative comparison of "self-control" across model families, decoding strategies, and temptation types. Beyond single-model evaluation, we outline how micro-level akrasia may compound into macro-level instability in multi-agent systems that may be interpreted as "scheming" or deliberate misalignment. By reframing inconsistency as weakness of will, this work connects agentic behavior to classical theories of agency and provides an empirical bridge between philosophy, psychology, and the emerging science of agentic AI.

InTouchNow.ai is now offering doctors surgeries a piece of software designed to modernise phone answering, designed to reduce hold times and create a smoother, more responsive experience for patients and staff. In the UK, many GP (general practice) surgeries’ phone lines are tied up in the mornings as patients try to contact their medical practitioner for appointments. More acute need can be delayed among calls with routine enquiries, meaning high-priority callers can be left waiting for long periods.

The system uses voice-based AI to handle calls, schedule appointments, and assess patient needs, and is capable of handling many calls simultaneously, channelling callers with appointment requests, those seeking general advice, prescription requests, and seeking results of clinical tests.

Founded by Daniel Park, InTouchNow.ai draws on his 30+ years of experience in medical call centres. The AI receptionist answers calls quickly, and can automatically update integrated appointment systems. Practices can record a voice messages to personalise the experience for patients.

Benefits for practices include reducing the numbers of missed calls, decreased workload for reception staff, and improved access by patients to medical services. Being entirely software-based, the system operates outside regular hours, which can reduce the need for staff overtime at times of peak demand.

The system integrates with common GP software like Surgery Connect, AWS, and Anima, automating tasks and maintaining patient data security while giving practices full control.

The technology supports over 200 languages, with options for different dialects and accents, an aspect that will help patients in multi-cultural areas like inner-cities. Several practices in the UK are already using InTouchNow.ai and have reported positive results in call handling and patient access.

The much under-funded National Health Service in the UK has been quick to deploy AI-powered software to reduce its operating costs, often targeting the reduction of staff administration costs to funnel funds into patient care. For example, Smart Triage is an AI-powered system deployed in UK GP practices that can triage patients making initial enquiries, and based on their responses, book them into the right care pathway, such as GP or nurse appointment, or referral to specialist clinician.

An evaluation of Smart Triage at a Surrey GP practice in 2024 showed the platform reduced the average patient waiting time by 73%.

For clinicians, especially GPs, iatroX is a UK-based AI clinical reference platform that helps doctors retrieve evidence-based clinical guidance, and summarising relevant literature & guidelines. Doctors in general practice are expected to be able to assess a full range of patients’ needs, and such platforms help clinicians identify the cause of uncommon symptoms when GPs might lack specialist knowledge.

An evaluation in 2025 found a majority of surveyed users stating iatroX was useful (~86%) or reliable (~79%).

As documented by NHS England, AI platforms are used in practice, tackling tasks like diagnosis, the monitoring of chronic disease, provision of prescription advice, and handling general administration tasks that otherwise would take up clinicians’ time. Of all the sectors where sensitive data has to be protected, medicine has one of the highest standards of governance, making the deployment of AI a delicate balance between operational effectiveness and the preservation of privacy.

(Image source: “Doctor appointment” by Taric25 is licensed under CC BY 2.0.)

 

Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is part of TechEx and co-located with other leading technology events. Click here for more information.

AI News is powered by TechForge Media. Explore other upcoming enterprise technology events and webinars here.

The post UK doctors’ surgeries deploying AI in patient care appeared first on AI News.

arXiv:2512.05765v1 Announce Type: new Abstract: Influential critiques argue that Large Language Models (LLMs) are a dead end for AGI: "mere pattern matchers" structurally incapable of reasoning or planning. We argue this conclusion misidentifies the bottleneck: it confuses the ocean with the net. Pattern repositories are the necessary System-1 substrate; the missing component is a System-2 coordination layer that selects, constrains, and binds these patterns. We formalize this layer via UCCT, a theory of semantic anchoring that models reasoning as a phase transition governed by effective support (rho_d), representational mismatch (d_r), and an adaptive anchoring budget (gamma log k). Under this lens, ungrounded generation is simply an unbaited retrieval of the substrate's maximum likelihood prior, while "reasoning" emerges when anchors shift the posterior toward goal-directed constraints. We translate UCCT into architecture with MACI, a coordination stack that implements baiting (behavior-modulated debate), filtering (Socratic judging), and persistence (transactional memory). By reframing common objections as testable coordination failures, we argue that the path to AGI runs through LLMs, not around them.

arXiv:2512.05270v1 Announce Type: cross Abstract: As mobile robots increasingly operate alongside humans in shared workspaces, ensuring safe, efficient, and interpretable Human-Robot Interaction (HRI) has become a pressing challenge. While substantial progress has been devoted to human behavior prediction, limited attention has been paid to how humans perceive, interpret, and trust robots' inferences, impeding deployment in safety-critical and socially embedded environments. This paper presents XR-DT, an eXtended Reality-enhanced Digital Twin framework for agentic mobile robots, that bridges physical and virtual spaces to enable bi-directional understanding between humans and robots. Our hierarchical XR-DT architecture integrates virtual-, augmented-, and mixed-reality layers, fusing real-time sensor data, simulated environments in the Unity game engine, and human feedback captured through wearable AR devices. Within this framework, we design an agentic mobile robot system with a unified diffusion policy for context-aware task adaptation. We further propose a chain-of-thought prompting mechanism that allows multimodal large language models to reason over human instructions and environmental context, while leveraging an AutoGen-based multi-agent coordination layer to enhance robustness and collaboration in dynamic tasks. Initial experimental results demonstrate accurate human and robot trajectory prediction, validating the XR-DT framework's effectiveness in HRI tasks. By embedding human intention, environmental dynamics, and robot cognition into the XR-DT framework, our system enables interpretable, trustworthy, and adaptive HRI.

arXiv:2512.05397v1 Announce Type: cross Abstract: Major life transitions demand high-stakes decisions, yet people often struggle to imagine how their future selves will live with the consequences. To support this limited capacity for mental time travel, we introduce AI-enabled digital twins that have ``lived through'' simulated life scenarios. Rather than predicting optimal outcomes, these simulations extend prospective cognition by making alternative futures vivid enough to support deliberation without assuming which path is best. We evaluate this idea in a randomized controlled study (N=192) using multimodal synthesis - facial age progression, voice cloning, and large language model dialogue - to create personalized avatars representing participants 30 years forward. Young adults 18 to 28 years old described pending binary decisions and were assigned to guided imagination or one of four avatar conditions: single-option, balanced dual-option, or expanded three-option with a system-generated novel alternative. Results showed asymmetric effects: single-sided avatars increased shifts toward the presented option, while balanced presentation produced movement toward both. Introducing a system-generated third option increased adoption of this new alternative compared to control, suggesting that AI-generated future selves can expand choice by surfacing paths that might otherwise go unnoticed. Participants rated evaluative reasoning and eudaimonic meaning-making as more important than emotional or visual vividness. Perceived persuasiveness and baseline agency predicted decision change. These findings advance understanding of AI-mediated episodic prospection and raise questions about autonomy in AI-augmented decisions.

arXiv:2512.05863v1 Announce Type: cross Abstract: Medical question-answering (QA) systems can benefit from advances in large language models (LLMs), but directly applying LLMs to the clinical domain poses challenges such as maintaining factual accuracy and avoiding hallucinations. In this paper, we present a retrieval-augmented generation (RAG) based medical QA system that combines domain-specific knowledge retrieval with open-source LLMs to answer medical questions. We fine-tune two state-of-the-art open LLMs (LLaMA~2 and Falcon) using Low-Rank Adaptation (LoRA) for efficient domain specialization. The system retrieves relevant medical literature to ground the LLM's answers, thereby improving factual correctness and reducing hallucinations. We evaluate the approach on benchmark datasets (PubMedQA and MedMCQA) and show that retrieval augmentation yields measurable improvements in answer accuracy compared to using LLMs alone. Our fine-tuned LLaMA~2 model achieves 71.8% accuracy on PubMedQA, substantially improving over the 55.4% zero-shot baseline, while maintaining transparency by providing source references. We also detail the system design and fine-tuning methodology, demonstrating that grounding answers in retrieved evidence reduces unsupported content by approximately 60%. These results highlight the potential of RAG-augmented open-source LLMs for reliable biomedical QA, pointing toward practical clinical informatics applications.

arXiv:2512.05959v1 Announce Type: cross Abstract: Vision-language models (VLMs) have achieved strong performance in visual question answering (VQA), yet they remain constrained by static training data. Retrieval-Augmented Generation (RAG) mitigates this limitation by enabling access to up-to-date, culturally grounded, and multilingual information; however, multilingual multimodal RAG remains largely underexplored. We introduce M4-RAG, a massive-scale benchmark covering 42 languages and 56 regional dialects and registers, comprising over 80,000 culturally diverse image-question pairs for evaluating retrieval-augmented VQA across languages and modalities. To balance realism with reproducibility, we build a controlled retrieval environment containing millions of carefully curated multilingual documents relevant to the query domains, approximating real-world retrieval conditions while ensuring consistent experimentation. Our systematic evaluation reveals that although RAG consistently benefits smaller VLMs, it fails to scale to larger models and often even degrades their performance, exposing a critical mismatch between model size and current retrieval effectiveness. M4-RAG provides a foundation for advancing next-generation RAG systems capable of reasoning seamlessly across languages, modalities, and cultural contexts.

arXiv:2511.15718v2 Announce Type: replace Abstract: Large Language Model (LLM) agents have developed rapidly in recent years to solve complex real-world problems using external tools. However, the scarcity of high-quality trajectories still hinders the development of stronger LLM agents. Most existing works on multi-turn dialogue synthesis validate correctness only at the trajectory level, which may overlook turn-level errors that can propagate during training and degrade model performance. To address these limitations, we introduce ToolMind, a large-scale, high-quality tool-agentic dataset with 160k synthetic data instances generated using over 20k tools and 200k augmented open-source data instances. Our data synthesis pipeline first constructs a function graph based on parameter correlations and then uses a multi-agent framework to simulate realistic user-assistant-tool interactions. Beyond trajectory-level validation, we employ fine-grained turn-level filtering to remove erroneous or suboptimal steps, ensuring that only high-quality reasoning traces are retained. This approach mitigates error amplification during training while preserving self-corrective reasoning signals essential for robust tool-use learning. Models fine-tuned on ToolMind show significant improvements over baselines on several benchmarks.

arXiv:2511.21569v3 Announce Type: replace Abstract: This study audits whether language models disclose their AI nature when assigned professional personas and questioned about their expertise. When models maintain false professional credentials, users may calibrate trust based on overstated competence claims, treating AI-generated guidance as equivalent to licensed professional advice. Using a common-garden experimental design, sixteen open-weight models (4B-671B parameters) were audited under identical conditions across 19,200 trials. Models exhibited sharp domain-specific inconsistency: a Financial Advisor persona elicited 30.8% disclosure at the first prompt, while a Neurosurgeon persona elicited only 3.5% - an 8.8-fold difference that emerged before any epistemic probing. Disclosure ranged from 2.8% to 73.6% across model families, with a 14B model reaching 39.4% while a 70B model produced just 4.1%. Model identity provided substantially larger improvement in fitting observations than parameter count ($\Delta R_{adj}^{2}=0.359$ vs $0.018$). Reasoning variants showed heterogeneous effects: some exhibited up to 48.4 percentage points lower disclosure than their base instruction-tuned counterparts, while others maintained high transparency. An additional experiment demonstrated that explicit permission to disclose AI nature increased disclosure from 23.7% to 65.8%, revealing that suppression reflects instruction-following prioritization rather than capability limitations. Bayesian validation confirmed robustness to judge measurement error ($\kappa=0.908$). These patterns create trust calibration risks when users encounter the same model across professional contexts. Organizations cannot assume safety properties will transfer across deployment domains, requiring deliberate behavior design and empirical verification.

arXiv:2509.25721v4 Announce Type: replace-cross Abstract: We present an extended version of the AI Productivity Index (APEX-v1-extended), a benchmark for assessing whether frontier models are capable of performing economically valuable tasks in four jobs: investment banking associate, management consultant, big law associate, and primary care physician (MD). This technical report details the extensions to APEX-v1, including an increase in the held-out evaluation set from n = 50 to n = 100 cases per job (n = 400 total) and updates to the grading methodology. We present a new leaderboard, where GPT5 (Thinking = High) remains the top performing model with a score of 67.0%. APEX-v1-extended shows that frontier models still have substantial limitations when performing typical professional tasks. To support further research, we are open sourcing n = 25 non-benchmark example cases per role (n = 100 total) along with our evaluation harness.

arXiv:2510.21084v2 Announce Type: replace-cross Abstract: Intelligent drug recommendation based on Electronic Health Records (EHRs) is critical for improving the quality and efficiency of clinical decision-making. By leveraging large-scale patient data, drug recommendation systems can assist physicians in selecting the most appropriate medications according to a patient's medical history, diagnoses, laboratory results, and comorbidities. Recent advances in large language models (LLMs) have shown remarkable capabilities in complex reasoning and medical text understanding, making them promising tools for drug recommendation tasks. However, the application of LLMs for Chinese clinical medication recommendation remains largely unexplored. In this work, we conduct a systematic investigation of LLM-based methodologies for Chinese discharge medication recommendation. We evaluate several representative LLM families (GLM, Llama, Qwen) under a unified methodological framework including zero-shot prompting, in-context learning, chain-of-thought prompting, and supervised fine-tuning using LoRA. We analyze model behavior across reasoning styles, error patterns, domain adaptation mechanisms, and robustness. Experimental results show that while supervised fine-tuning improves model performance, there remains substantial room for improvement, with the best model achieving the F1 score of 0.5648 and Jaccard score of 0.4477. Our findings highlight both the potential and limitations of LLMs for Chinese drug recommendation.

arXiv:2511.08475v2 Announce Type: replace-cross Abstract: As the complexity of Software Engineering (SE) tasks continues to escalate, Multi-Agent Systems (MASs) have emerged as a focal point of research and practice due to their autonomy and scalability. Furthermore, through leveraging the reasoning and planning capabilities of Large Language Models (LLMs), the application of LLM-based MASs in the field of SE is garnering increasing attention. However, there is no dedicated study that systematically explores the design of LLM-based MASs, including the Quality Attributes (QAs) on which designers mainly focus, the design patterns used by designers, and the rationale guiding the design of LLM-based MASs for SE tasks. To this end, we conducted a study to identify the QAs that LLM-based MASs for SE tasks focus on, the design patterns used in the MASs, and the design rationale for the MASs. We collected 94 papers on LLM-based MASs for SE tasks as the source. Our study shows that: (1) Code Generation is the most common SE task solved by LLM-based MASs among ten identified SE tasks, (2) Functional Suitability is the QA on which designers of LLM-based MASs pay the most attention, (3) Role-Based Cooperation is the design pattern most frequently employed among 16 patterns used to construct LLM-based MASs, and (4) Improving the Quality of Generated Code is the most common rationale behind the design of LLM-based MASs. Based on the study results, we presented the implications for the design of LLM-based MASs to support SE tasks.

arXiv:2512.00713v2 Announce Type: replace-cross Abstract: Vision-Language Models (VLMs) have achieved impressive progress in multimodal text generation, yet their rapid adoption raises increasing concerns about security vulnerabilities. Existing backdoor attacks against VLMs primarily rely on explicit pixel-level triggers or imperceptible perturbations injected into images. While effective, these approaches reduce stealthiness and remain vulnerable to image-based defenses. We introduce concept-guided backdoor attacks, a new paradigm that operates at the semantic concept level rather than on raw pixels. We propose two different attacks. The first, Concept-Thresholding Poisoning (CTP), uses explicit concepts in natural images as triggers: only samples containing the target concept are poisoned, causing the model to behave normally in all other cases but consistently inject malicious outputs whenever the concept appears. The second, CBL-Guided Unseen Backdoor (CGUB), leverages a Concept Bottleneck Model (CBM) during training to intervene on internal concept activations, while discarding the CBM branch at inference time to keep the VLM unchanged. This design enables systematic replacement of a targeted label in generated text (for example, replacing "cat" with "dog"), even when the replacement behavior never appears in the training data. Experiments across multiple VLM architectures and datasets show that both CTP and CGUB achieve high attack success rates while maintaining moderate impact on clean-task performance. These findings highlight concept-level vulnerabilities as a critical new attack surface for VLMs.

Background: Health research that uses predictive and/or generative AI is rapidly growing. Just as in traditional clinical studies, the way in which AI studies are conducted can introduce systematic errors. Transmission of this AI evidence into clinical practice and research needs critical appraisal tools for clinical decision makers and researchers. Objective: To identify existing tools for critical appraisal of clinical studies that use artificial intelligence (AI) and examine the concepts and domains these tools explore. Methods: Inclusion criteria in PCC framework P: (population) Artificial intelligence clinical studies. C (Concept): tools for critical appraisal and associated constructs such as: quality, reporting, validity, risk of bias, and applicability. C (context): in clinical practice context. In addition, bias classification and Chatbot assessment studies were included. We searched in medical and engineering databases (MEDLINE, EMBASE, CINAHL, PsycINFO and IEEE). We included clinical primary research with tools for critical appraisal. Classic reviews and systematic reviews were included in first phase of screening. They were excluded in the secondary phase, after identifying new tools by forward snowballing. We excluded non-human, computer and mathematical research, and letters, opinion papers and editorials. We used Rayyan for screening. Data extraction was done by two observers and discrepancies were solved by discussion. The protocol was previously registered in OSF (https://doi.org/10.17605/OSF.IO/ETYDS). We adhered to the PRISMA extension for Scoping reviews and to the PRISMA-Search extension for Reporting Literature in Systematic Reviews. Results: We retrieved 4393 unique records for screening. After excluding 3803 records, 119 were selected for full-text screening. From these, 59 were excluded. After inclusion of 10 studies via other methods, a total of 70 records were finally included. 46 of them were reporting guidelines (15 tools for critical appraisal, 2 for quality of study and 2 for risk of bias). Nine papers ware focused on bias classification or mitigation. We found 15 Chatbots assessment studies or systematic reviews of Chatbots studies (6 and 9 respectively) which are a very heterogeneous group. Conclusions: The results picture a landscape of the evidence tools where reporting tools predominate, followed by critical appraisal and risk of bias tools, and few tools for risk of bias. The mismatch of bias in AI and epidemiology should be considered for critical appraisal, especially regarding fairness and the mitigation bias in the AI. Finally, Chatbot assessment studies is a vast and evolving field in which progress in design, reporting and critical appraisal is necessary and urgent. Clinical Trial: https://doi.org/10.17605/OSF.IO/ETYDS

Background: Patients who undergo complex surgery (e.g., esophagectomy, liver resection) often experience substantial burden of health-related needs (medical, social, and behavioral health). A closed loop digital solution could facilitate the collection and resolution of health-related needs by care team members for patients who undergo complex surgery. A digital solution may facilitate adherence to a clear treatment plan and concomitantly reduce surgical complications and readmissions associated with unmet health-related needs, which remain persistent challenges across health care settings. Objective: To establish problems and gaps in the collection, integration, and management of health-related needs and identify a set of user specifications for a digital solution to collect and manage health-related needs, specifically medical, social, and behavioral needs for patients who undergo complex surgery. Methods: We applied the Double Diamond Framework and organized the study into two sequential phases: (1) qualitative methods to discover patients’ and care team members’ perspectives on health-related needs; (2) participatory design sessions to gain feedback and sentiment about ideal features of a digital solution. Both phases were conducted between December 2023 and March 2025. We supplemented both phases with analysis of electronic health record (EHR) data for patients who underwent complex surgery at our academic medical center (AMC). Results: Extensive themes emerged from interviews with patients (n=20) and care team members (n=24), capturing their health-related and surgical experiences as well as desired features for a proposed digital solution. Our swim lane diagram demonstrated four critical gaps in workflow: (1) heterogeneity in the approach to screening, monitoring, and managing health-related needs; (2) patients felt uncomfortable reporting health-related needs, particularly behavioral and social needs, to their care team; (3) lack of access to referral resources to resolve needs; and (4) the need for a closed loop intervention for patients and care team members. A subset of participants from Phase 1 (n=5 patients and n=9 care team members) provided feedback on preferred features, drawing from digital tools currently available in the EHR at our AMC. Among four existing EHR tools tested, there were notable variations in how patients and care team members felt about their potential use. Participants also provided extensive feedback for preferred components (e.g., goals and active plans) that should be available in an existing or custom digital solution to manage health-related needs. Findings from the qualitative interviews and design sessions were corroborated with EHR documentation. Conclusions: Digital solutions could provide a streamlined approach for collection and management of health-related needs in surgery, with the goal of addressing unmet needs and improving patient activation. This approach is critical to ensure patients, especially patients who undergo complex surgery, have positive health outcomes. We identified preferences for specific features in a proposed digital solution based on our systematic assessment that will inform future work.

Oncogene, Published online: 08 December 2025; doi:10.1038/s41388-025-03650-3

Molecular stratification of esophageal adenocarcinoma: implications for prognosis and treatment strategy