Category: Daily AI News

PRISMA: Reinforcement Learning Guided Two-Stage Policy Optimization in Multi-Agent Architecture for Open-Domain Multi-Hop Question Answeringcs.AI updates on arXiv.org arXiv:2601.05465v1 Announce Type: new
Abstract: Answering real-world open-domain multi-hop questions over massive corpora is a critical challenge in Retrieval-Augmented Generation (RAG) systems. Recent research employs reinforcement learning (RL) to end-to-end optimize the retrieval-augmented reasoning process, directly enhancing its capacity to resolve complex queries. However, reliable deployment is hindered by two obstacles. 1) Retrieval Collapse: iterative retrieval over large corpora fails to locate intermediate evidence containing bridge answers without reasoning-guided planning, causing downstream reasoning to collapse. 2) Learning Instability: end-to-end trajectory training suffers from weak credit assignment across reasoning chains and poor error localization across modules, causing overfitting to benchmark-specific heuristics that limit transferability and stability. To address these problems, we propose PRISMA, a decoupled RL-guided framework featuring a Plan-Retrieve-Inspect-Solve-Memoize architecture. PRISMA’s strength lies in reasoning-guided collaboration: the Inspector provides reasoning-based feedback to refine the Planner’s decomposition and fine-grained retrieval, while enforcing evidence-grounded reasoning in the Solver. We optimize individual agent capabilities via Two-Stage Group Relative Policy Optimization (GRPO). Stage I calibrates the Planner and Solver as specialized experts in planning and reasoning, while Stage II utilizes Observation-Aware Residual Policy Optimization (OARPO) to enhance the Inspector’s ability to verify context and trigger targeted recovery. Experiments show that PRISMA achieves state-of-the-art performance on ten benchmarks and can be deployed efficiently in real-world scenarios.

 arXiv:2601.05465v1 Announce Type: new
Abstract: Answering real-world open-domain multi-hop questions over massive corpora is a critical challenge in Retrieval-Augmented Generation (RAG) systems. Recent research employs reinforcement learning (RL) to end-to-end optimize the retrieval-augmented reasoning process, directly enhancing its capacity to resolve complex queries. However, reliable deployment is hindered by two obstacles. 1) Retrieval Collapse: iterative retrieval over large corpora fails to locate intermediate evidence containing bridge answers without reasoning-guided planning, causing downstream reasoning to collapse. 2) Learning Instability: end-to-end trajectory training suffers from weak credit assignment across reasoning chains and poor error localization across modules, causing overfitting to benchmark-specific heuristics that limit transferability and stability. To address these problems, we propose PRISMA, a decoupled RL-guided framework featuring a Plan-Retrieve-Inspect-Solve-Memoize architecture. PRISMA’s strength lies in reasoning-guided collaboration: the Inspector provides reasoning-based feedback to refine the Planner’s decomposition and fine-grained retrieval, while enforcing evidence-grounded reasoning in the Solver. We optimize individual agent capabilities via Two-Stage Group Relative Policy Optimization (GRPO). Stage I calibrates the Planner and Solver as specialized experts in planning and reasoning, while Stage II utilizes Observation-Aware Residual Policy Optimization (OARPO) to enhance the Inspector’s ability to verify context and trigger targeted recovery. Experiments show that PRISMA achieves state-of-the-art performance on ten benchmarks and can be deployed efficiently in real-world scenarios. Read More  

Reasoning Models Will Blatantly Lie About Their Reasoningcs.AI updates on arXiv.org arXiv:2601.07663v2 Announce Type: replace
Abstract: It has been shown that Large Reasoning Models (LRMs) may not *say what they think*: they do not always volunteer information about how certain parts of the input influence their reasoning. But it is one thing for a model to *omit* such information and another, worse thing to *lie* about it. Here, we extend the work of Chen et al. (2025) to show that LRMs will do just this: they will flatly deny relying on hints provided in the prompt in answering multiple choice questions — even when directly asked to reflect on unusual (i.e. hinted) prompt content, even when allowed to use hints, and even though experiments *show* them to be using the hints. Our results thus have discouraging implications for CoT monitoring and interpretability.

 arXiv:2601.07663v2 Announce Type: replace
Abstract: It has been shown that Large Reasoning Models (LRMs) may not *say what they think*: they do not always volunteer information about how certain parts of the input influence their reasoning. But it is one thing for a model to *omit* such information and another, worse thing to *lie* about it. Here, we extend the work of Chen et al. (2025) to show that LRMs will do just this: they will flatly deny relying on hints provided in the prompt in answering multiple choice questions — even when directly asked to reflect on unusual (i.e. hinted) prompt content, even when allowed to use hints, and even though experiments *show* them to be using the hints. Our results thus have discouraging implications for CoT monitoring and interpretability. Read More  

7 AI Automation Tools for Streamlined WorkflowsKDnuggets This list focuses on tools that streamline real workflows across data, operations, and content, not flashy demos or brittle bots. Each one earns its place by reducing manual effort while keeping humans in the loop where it actually matters.

 This list focuses on tools that streamline real workflows across data, operations, and content, not flashy demos or brittle bots. Each one earns its place by reducing manual effort while keeping humans in the loop where it actually matters. Read More  

Do You Smell That? Hidden Technical Debt in AI DevelopmentTowards Data Science Why speed without standards creates fragile AI products
The post Do You Smell That? Hidden Technical Debt in AI Development appeared first on Towards Data Science.

 Why speed without standards creates fragile AI products
The post Do You Smell That? Hidden Technical Debt in AI Development appeared first on Towards Data Science. Read More  

McKinsey tests AI chatbot in early stages of graduate recruitmentAI News Hiring at large firms has long relied on interviews, tests, and human judgment. That process is starting to shift. McKinsey has begun using an AI chatbot as part of its graduate recruitment process, signalling a shift in how professional services organisations evaluate early-career candidates. The chatbot is being used during the initial stages of recruitment,
The post McKinsey tests AI chatbot in early stages of graduate recruitment appeared first on AI News.

 Hiring at large firms has long relied on interviews, tests, and human judgment. That process is starting to shift. McKinsey has begun using an AI chatbot as part of its graduate recruitment process, signalling a shift in how professional services organisations evaluate early-career candidates. The chatbot is being used during the initial stages of recruitment,
The post McKinsey tests AI chatbot in early stages of graduate recruitment appeared first on AI News. Read More  

AI medical diagnostics race intensifies as OpenAI, Google, and Anthropic launch competing healthcare toolsAI News OpenAI, Google, and Anthropic announced specialised medical AI capabilities within days of each other this month, a clustering that suggests competitive pressure rather than coincidental timing. Yet none of the releases are cleared as medical devices, approved for clinical use, or available for direct patient diagnosis—despite marketing language emphasising healthcare transformation. OpenAI introduced ChatGPT Health on January
The post AI medical diagnostics race intensifies as OpenAI, Google, and Anthropic launch competing healthcare tools appeared first on AI News.

 OpenAI, Google, and Anthropic announced specialised medical AI capabilities within days of each other this month, a clustering that suggests competitive pressure rather than coincidental timing. Yet none of the releases are cleared as medical devices, approved for clinical use, or available for direct patient diagnosis—despite marketing language emphasising healthcare transformation. OpenAI introduced ChatGPT Health on January
The post AI medical diagnostics race intensifies as OpenAI, Google, and Anthropic launch competing healthcare tools appeared first on AI News. Read More  

Using Subgraph GNNs for Node Classification:an Overlooked Potential Approachcs.AI updates on arXiv.org arXiv:2503.06614v2 Announce Type: replace-cross
Abstract: Previous studies have demonstrated the strong performance of Graph Neural Networks (GNNs) in node classification. However, most existing GNNs adopt a node-centric perspective and rely on global message passing, leading to high computational and memory costs that hinder scalability. To mitigate these challenges, subgraph-based methods have been introduced, leveraging local subgraphs as approximations of full computational trees. While this approach improves efficiency, it often suffers from performance degradation due to the loss of global contextual information, limiting its effectiveness compared to global GNNs. To address this trade-off between scalability and classification accuracy, we reformulate the node classification task as a subgraph classification problem and propose SubGND (Subgraph GNN for NoDe). This framework introduces a differentiated zero-padding strategy and an Ego-Alter subgraph representation method to resolve label conflicts while incorporating an Adaptive Feature Scaling Mechanism to dynamically adjust feature contributions based on dataset-specific dependencies. Experimental results on six benchmark datasets demonstrate that SubGND achieves performance comparable to or surpassing global message-passing GNNs, particularly in heterophilic settings, highlighting its effectiveness and scalability as a promising solution for node classification.

 arXiv:2503.06614v2 Announce Type: replace-cross
Abstract: Previous studies have demonstrated the strong performance of Graph Neural Networks (GNNs) in node classification. However, most existing GNNs adopt a node-centric perspective and rely on global message passing, leading to high computational and memory costs that hinder scalability. To mitigate these challenges, subgraph-based methods have been introduced, leveraging local subgraphs as approximations of full computational trees. While this approach improves efficiency, it often suffers from performance degradation due to the loss of global contextual information, limiting its effectiveness compared to global GNNs. To address this trade-off between scalability and classification accuracy, we reformulate the node classification task as a subgraph classification problem and propose SubGND (Subgraph GNN for NoDe). This framework introduces a differentiated zero-padding strategy and an Ego-Alter subgraph representation method to resolve label conflicts while incorporating an Adaptive Feature Scaling Mechanism to dynamically adjust feature contributions based on dataset-specific dependencies. Experimental results on six benchmark datasets demonstrate that SubGND achieves performance comparable to or surpassing global message-passing GNNs, particularly in heterophilic settings, highlighting its effectiveness and scalability as a promising solution for node classification. Read More  

Meeting the new ETSI standard for AI securityAI News The ETSI EN 304 223 standard introduces baseline security requirements for AI that enterprises must integrate into governance frameworks. As organisations embed machine learning into their core operations, this European Standard (EN) establishes concrete provisions for securing AI models and systems. It stands as the first globally applicable European Standard for AI cybersecurity, having secured
The post Meeting the new ETSI standard for AI security appeared first on AI News.

 The ETSI EN 304 223 standard introduces baseline security requirements for AI that enterprises must integrate into governance frameworks. As organisations embed machine learning into their core operations, this European Standard (EN) establishes concrete provisions for securing AI models and systems. It stands as the first globally applicable European Standard for AI cybersecurity, having secured
The post Meeting the new ETSI standard for AI security appeared first on AI News. Read More  

Debiasing Large Language Models via Adaptive Causal Prompting with Sketch-of-Thoughtcs.AI updates on arXiv.org arXiv:2601.08108v1 Announce Type: cross
Abstract: Despite notable advancements in prompting methods for Large Language Models (LLMs), such as Chain-of-Thought (CoT), existing strategies still suffer from excessive token usage and limited generalisability across diverse reasoning tasks. To address these limitations, we propose an Adaptive Causal Prompting with Sketch-of-Thought (ACPS) framework, which leverages structural causal models to infer the causal effect of a query on its answer and adaptively select an appropriate intervention (i.e., standard front-door and conditional front-door adjustments). This design enables generalisable causal reasoning across heterogeneous tasks without task-specific retraining. By replacing verbose CoT with concise Sketch-of-Thought, ACPS enables efficient reasoning that significantly reduces token usage and inference cost. Extensive experiments on multiple reasoning benchmarks and LLMs demonstrate that ACPS consistently outperforms existing prompting baselines in terms of accuracy, robustness, and computational efficiency.

 arXiv:2601.08108v1 Announce Type: cross
Abstract: Despite notable advancements in prompting methods for Large Language Models (LLMs), such as Chain-of-Thought (CoT), existing strategies still suffer from excessive token usage and limited generalisability across diverse reasoning tasks. To address these limitations, we propose an Adaptive Causal Prompting with Sketch-of-Thought (ACPS) framework, which leverages structural causal models to infer the causal effect of a query on its answer and adaptively select an appropriate intervention (i.e., standard front-door and conditional front-door adjustments). This design enables generalisable causal reasoning across heterogeneous tasks without task-specific retraining. By replacing verbose CoT with concise Sketch-of-Thought, ACPS enables efficient reasoning that significantly reduces token usage and inference cost. Extensive experiments on multiple reasoning benchmarks and LLMs demonstrate that ACPS consistently outperforms existing prompting baselines in terms of accuracy, robustness, and computational efficiency. Read More  

Enhancing Large Language Models for Time-Series Forecasting via Vector-Injected In-Context Learningcs.AI updates on arXiv.org arXiv:2601.07903v2 Announce Type: cross
Abstract: The World Wide Web needs reliable predictive capabilities to respond to changes in user behavior and usage patterns. Time series forecasting (TSF) is a key means to achieve this goal. In recent years, the large language models (LLMs) for TSF (LLM4TSF) have achieved good performance. However, there is a significant difference between pretraining corpora and time series data, making it hard to guarantee forecasting quality when directly applying LLMs to TSF; fine-tuning LLMs can mitigate this issue, but often incurs substantial computational overhead. Thus, LLM4TSF faces a dual challenge of prediction performance and compute overhead. To address this, we aim to explore a method for improving the forecasting performance of LLM4TSF while freezing all LLM parameters to reduce computational overhead. Inspired by in-context learning (ICL), we propose LVICL. LVICL uses our vector-injected ICL to inject example information into a frozen LLM, eliciting its in-context learning ability and thereby enhancing its performance on the example-related task (i.e., TSF). Specifically, we first use the LLM together with a learnable context vector adapter to extract a context vector from multiple examples adaptively. This vector contains compressed, example-related information. Subsequently, during the forward pass, we inject this vector into every layer of the LLM to improve forecasting performance. Compared with conventional ICL that adds examples into the prompt, our vector-injected ICL does not increase prompt length; moreover, adaptively deriving a context vector from examples suppresses components harmful to forecasting, thereby improving model performance. Extensive experiments demonstrate the effectiveness of our approach.

 arXiv:2601.07903v2 Announce Type: cross
Abstract: The World Wide Web needs reliable predictive capabilities to respond to changes in user behavior and usage patterns. Time series forecasting (TSF) is a key means to achieve this goal. In recent years, the large language models (LLMs) for TSF (LLM4TSF) have achieved good performance. However, there is a significant difference between pretraining corpora and time series data, making it hard to guarantee forecasting quality when directly applying LLMs to TSF; fine-tuning LLMs can mitigate this issue, but often incurs substantial computational overhead. Thus, LLM4TSF faces a dual challenge of prediction performance and compute overhead. To address this, we aim to explore a method for improving the forecasting performance of LLM4TSF while freezing all LLM parameters to reduce computational overhead. Inspired by in-context learning (ICL), we propose LVICL. LVICL uses our vector-injected ICL to inject example information into a frozen LLM, eliciting its in-context learning ability and thereby enhancing its performance on the example-related task (i.e., TSF). Specifically, we first use the LLM together with a learnable context vector adapter to extract a context vector from multiple examples adaptively. This vector contains compressed, example-related information. Subsequently, during the forward pass, we inject this vector into every layer of the LLM to improve forecasting performance. Compared with conventional ICL that adds examples into the prompt, our vector-injected ICL does not increase prompt length; moreover, adaptively deriving a context vector from examples suppresses components harmful to forecasting, thereby improving model performance. Extensive experiments demonstrate the effectiveness of our approach. Read More