Beyond the Black Box: Building Trustworthy AI
A new approach argues that current artificial intelligence needs a grounding in verifiable world models and shared understanding to ensure safety and reliability.
Science
Turning Warnings into Action: AI for Proactive Disaster Resilience
A new system leverages generative AI to move beyond simply alerting communities to providing tailored, actionable recommendations that improve preparedness and response.
Keeping Critical Medical Equipment Running, Even When Experts Aren’t Nearby
A new AI-powered platform aims to empower biomedical technicians in under-resourced regions with the tools they need to diagnose, maintain, and repair vital medical devices.
What Audio Models ‘Hear’: Bridging the Gap Between AI and the Human Brain
New research explores whether the internal representations of advanced audio AI models align with the way the human brain processes sound during natural listening.
Smart Edges: Balancing Wireless and Compute for Real-Time Video
![The system’s ergodic capacity is demonstrably affected by parameters including power control ε, receiver antenna count [latex]M[/latex], base station density [latex]\lambda_b[/latex], and reuse factor δ, with performance fluctuating based on the distribution of user distance and maximum distance to the serving base station [latex]r_{max}[/latex].](https://arxiv.org/html/2601.16848v1/x7.png)
This review details a new mathematical approach to efficiently allocate resources in edge computing networks, ensuring reliable performance for demanding applications like video analytics.
Can Your Voice Reveal Your Mental Wellbeing?
A new study explores how artificial intelligence can analyze vocal patterns to assess mental stability with improved accuracy.
Seeing Around Corners: Predicting V2X Network Performance in Real Time
![The study leverages ray tracing to model light propagation across a road segment on the Ookayama campus of the Institute of Science Tokyo, Japan, demonstrating a representation-VaN3Twin-capable of accurately simulating realistic visual environments for experimentation and analysis of light behavior [latex] [/latex].](https://arxiv.org/html/2601.16559v1/figures/sec_4/rays_temp.png)
Researchers are leveraging digital twin technology and advanced ray tracing to forecast V2X communication conditions before they occur, enabling proactive network optimization.
Decoding Bug Reports: A Linguistic Approach to Concurrency Errors
![Linguistic patterns, when subjected to cross-dataset analysis, demonstrate a surprising capacity to generalize-a phenomenon suggesting underlying structural consistencies that transcend superficial variations in data representation and hinting at a universal grammar governing language itself, potentially describable through [latex] \mathcal{L}(x, y) [/latex].](https://arxiv.org/html/2601.16338v1/x14.png)
Researchers are harnessing the power of language to automatically identify bug reports stemming from tricky concurrency issues in software.
Fortifying the Network: A Strategic Approach to Cybersecurity
![The study demonstrates that strategic allocation of defensive resources-specifically, an optimized strategy [latex]\bm{q}^{\*}[/latex]-effectively curtails the spread of contagion compared to scenarios with no protection or random budget distribution, as evidenced by both the time evolution of infection size and the resulting asymptotic infection levels relative to the defender’s budget.](https://arxiv.org/html/2601.16805v1/x12.png)
New research details a framework for optimizing security investments by modeling the dynamic interplay between attackers and defenders in interconnected systems.
Beyond Patchwork: Why Software Engineering Research Needs a Systems Reset
A new review argues that incremental improvements to software engineering research are failing to address fundamental problems within the field’s evaluation and publication practices.