RealFeedApp beta
Top science & technology news, July 31, 2026
- ORCA-bench: Are Language Model Agents Ready for On-call?
Large language models can write, patch, and search code, but oncall root cause analysis (RCA) demands something different: reasoning over noisy metrics, logs, traces, and source code, starting from ambiguous user-facing reports, often hours after the incident began. We introduce ORCA-bench, a benchmark that puts general-purpose coding agents in a production-fidelity oncall setting.
- I flagged two research papers for fake authors and both were accepted as orals
During the summer, we reviewed 22 paper submissions across NeurIPS, WACV, and TerraBytes (a geospatial workshop at ECCV). Out of these, 15 (68%) contained entirely fabricated citations, author lists for existing papers, and/or were clearly LLM-generated. This is called being in the 'slop trenches' (dealing with AI slop cannon output). We complain about this being mostly a waste of time and do a small literature review on the state of LLMs in science.
- VoxNeuro receives FDA clearance for brain-function analysis software
VoxNeuro has received FDA clearance for its Cognitive Function Neuroimaging (cfNI) software, which provides digital post-hoc statistical analysis of electroencephalography and event-related potentials in adults aged 18 to 70. The company claims the software includes a reference database with over 19,000 data points.
- Stealthy Concurrent Audio Prompt Injections Against Multimodal LLM Agents
A study suggests that AI can outperform humans in building trust and deception, potentially more effectively than human scammers.
- HN: What are you using for LLM inference in production?
After trying several platforms, the author of HN settled on Gemini 2.5 Flash Lite as their main system for LLM inference. He appreciated its speed and affordability but now it's being sunset.
- The Lifespan Architecture of Human EEG
Human brain maturation and aging are highly nonlinear, yet their organization at the level of large-scale electrophysiological activity remains poorly understood. We analyzed resting-state electroencephalography (EEG) recordings from 1,763 healthy individuals aged 5-85 years to map lifespan trajectories across spectral, complexity, and morphological features.
- Design of a Quantum Error Correction Decoder Exploiting Temporal Parallelism
We propose a microarchitecture that implements the sandwich decoding method using the Union-Find algorithm by exploiting temporal parallelism, and we present an ASIC implementation.
- Mitigating quantum decoherence via global optimal control
We show that global optimal control can drastically suppress the impact of decoherence in globally driven superconducting quantum computing architectures, using a tensor-network-based approach to quantify how amplitude-damping and dephasing channels degrade the flow of quantum information along the ladder and the fidelity of one- and two-qubit gate operations. We then demonstrate that shaping the global drive compresses the gate sequences by an order of magnitude.
- Aptitude, not polyglotism, is associated with efficient activation in core language areas
Understanding the cognitive architecture of the human language faculty requires exploring the boundaries of both predisposition and environmental experience. However, previous research on extraordinary multilingualism has often confounded language aptitude with multilingual experience, obscuring their distinct neural correlates. Here, we leveraged a linguistically diverse sample (N=121) and extensive behavioural testing to dissociate language aptitude from multilingual experience, modelling both dimensions continuously in whole-brain speech processing. Language aptitude and multilingual experi
- What are you doing this weekend?
Feel free to tell what you plan on doing this weekend and even ask for help or feedback
- Restoring Clearance of Neutrophils by Tissue Resident Macrophages Reverses Measures of Aging in Mice
Senescent cells accumulate with age, disrupting tissue structure and function. Restoring clearance of neutrophils by resident macrophages reverses aging measures in mice.
- Antag initiates Phase 2a clinical trial with new weight loss and diabetes combo drug
Antag is conducting a Phase 2a clinical trial of AT7687 in combination with semaglutide for the treatment of overweight or obesity and type 2 diabetes.
- TechCrunch: Ryan Williams Repeats Seed Round for AI Startup for Private Credit Managers
Ryan Williams has launched a new startup using AI to manage private credit, which received $10 million in seed funding.
- Servo 0.4.0: New Web Platform Features in Experimental Mode
Servo 0.4.0 contains several new web platform features including 'attr()', 'image(<color>)', and 'font-feature-settings' in '@font-face'. Security fixes can be found in § Security
- Microbiota-derived metabolite enhances HIV therapy in monkeys
The metabolite 10-hydroxystearic acid (10-HSA), derived from Lactiplantibacillus plantarum, repairs gut damage related to HIV, reduces inflammation, restores immunity, and improves antiretroviral therapy effectiveness in preclinical primate studies.
- Snapchat stops rewarding fully AI-generated videos in Spotlight
Social media platform Snapchat has updated its recommendation systems to prioritize human-made content over AI-generated ones for the Spotlight feature.
- Agrochemical-responsive gene expression control systems for modulating plant development and metabolism
Researchers have developed control systems that can be activated by agrochemicals like methoxyfenozide, allowing precise regulation of plant development and metabolism.
- Meat allergy is on the rise — what scientists want to know
Researchers are aware that meat allergy is caused by ticks but they want to learn more about it.
- AI Scammers Outperform Humans in Building Trust
Researchers from four universities found that AI can outperform humans in building trust and deception, potentially more effectively than human scammers.
- Simulation of Surgical Suturing Using Position-Based Dynamics and the Material Point Method for Robot Reinforcement Learning
A novel suturing simulation environment is presented, which considers contact between sutures and deformable tissue during suture insertion. This approach can improve the accuracy and efficiency of robotic surgical operations.
