All posts tagged: accuracy

Corti’s new Symphony for Speech-to-Text model beats OpenAI at medical terminology accuracy, highlighting the value of specialized AI

Corti’s new Symphony for Speech-to-Text model beats OpenAI at medical terminology accuracy, highlighting the value of specialized AI

Published by skeptic

Today, Copenhagen-based healthcare AI Corti is launching Symphony for Speech-to-Text, a new generation of clinical-grade speech recognition models engineered specifically for real-time dictation, conversational transcription, and batch audio processing — and their accuracy rate is the highest for this specific use case yet recorded. “We are focused on ensuring our AI scribes can be trusted by physicians, medical practitioners and patients…the entire healthcare system,” said Andreas Cleve, co-founder and CEO of Corti, in an exclusive video call interview with VentureBeat. The performance data the company is bringing to the table paints a stark picture of the current state of enterprise AI: when it comes to highly regulated, specialized industries, domain-specific models can beat out the foundation model providers. In a newly published research paper, Corti revealed that its new clinical-grade speech models reduced word error rates (WER) by up to 93% when compared against leading generalist speech models and APIs on medical terminology. On English medical terminology, its Symphony for Speech-to-Text achieved a remarkably low 1.4% WER. By comparison, OpenAI’s speech model registered a 17.7% …

RAG precision tuning can quietly cut retrieval accuracy by 40%, putting agentic pipelines at risk

RAG precision tuning can quietly cut retrieval accuracy by 40%, putting agentic pipelines at risk

Published by skeptic

Enterprise teams that fine-tune their RAG embedding models for better precision may be unintentionally degrading the retrieval quality those pipelines depend on, according to new research from Redis. The paper, “Training for Compositional Sensitivity Reduces Dense Retrieval Generalization,” tested what happens when teams train embedding models for compositional sensitivity. That is the ability to catch sentences that look nearly identical but mean something different — “the dog bit the man” versus “the man bit the dog,” or a negation flip that reverses a statement’s meaning entirely. That training consistently broke dense retrieval generalization, how well a model retrieves correctly across broad topics and domains it wasn’t specifically trained on. Performance dropped by 8 to 9 percent on smaller models and by 40 percent on a current mid-size embedding model teams are actively using in production. The findings have direct implications for enterprise teams building agentic AI pipelines, where retrieval quality determines what context flows into an agent’s reasoning chain. A retrieval error in a single-stage pipeline returns a wrong answer. The same error in an …

AI predicts early skin cancer risk with 73% accuracy

AI predicts early skin cancer risk with 73% accuracy

Published by skeptic

A study on the whole adult population of Sweden via analysing registry data on age, sex, diagnosis and socioeconomic status found that artificial intelligence (AI) models could predict rates of melanoma with almost 73% accuracy. A collaborative study between the University of Gothenburg and Chalmers University of Technology applied analytical AI models to a data pool containing over six million Swedish adults. Using registry data of 6,036,186 individuals and informed by demographic factors such as age and sex, the AI was able to identify small groups at high risk for developing melanoma with a high level of accuracy. The data focused on over 38,000 cancer diagnoses When only age and sex were factored into the registry data, the AI models were able to distinguish people who would later develop melanoma with roughly 64% accuracy. A more advanced model informed with comprehensive demographic data managed to increase accuracy to 73%. When informed with diagnoses, medications and sociodemographic data, the models could identify smaller high-risk groups, who faced a 33% risk of developing melanoma within five years. …

Your Answers To These 2 Questions Can Predict Divorce With Surprising Accuracy, Research Says

Your Answers To These 2 Questions Can Predict Divorce With Surprising Accuracy, Research Says

Published by Jenni Sidey

If you could ask yourself just two questions to determine the strength of your relationship, you would most certainly do it, right? Economics researchers Leora Friedberg and Steven Stern from the University of Virginia studied the answers from 3,597 couples with just two questions in mind, and it turns out they were accurately able to successfully predict which couples would head into divorce. They determined that partners who were able to answer the two posed questions in a positive light were more likely to stay together than those who couldn’t. Just what two questions did they ask? They seem pretty simplistic initially, but on a deeper level, they tap into core dynamics like how partners view each other and how they handle conflict. These patterns can be strong predictors of divorce, while answers reflecting respect and trust are linked to lasting relationships.  Your answers to these questions can predict divorce with surprising accuracy, research says: 1. How do you think your level of happiness would be different if you and your partner separated? This question …

TurboQuant Algorithm Lowers LLM Costs Without Accuracy Loss

TurboQuant Algorithm Lowers LLM Costs Without Accuracy Loss

Published by skeptic

Google’s TurboQuant is making waves in the AI hardware sector by addressing long-standing challenges in memory usage and processing efficiency. Developed with components like the Quantized Johnson-Lindenstrauss Algorithm, TurboQuant achieves up to sixfold reductions in memory requirements while preserving model accuracy. This compression algorithm also accelerates processing speeds by as much as eight times, allowing faster and more cost-effective deployment of large language models (LLMs). As Wes Roth explains, these advancements are reshaping how enterprises approach AI infrastructure, with significant implications for both operational efficiency and the broader hardware market. Explore how TurboQuant’s capabilities translate into practical benefits, from reducing inference costs by 50% to optimizing GPU utilization for existing hardware. Gain insight into its potential to extend context windows and support larger models, opening doors for more sophisticated AI applications. Additionally, understand the ripple effects on the memory chip market, where declining demand for high-capacity components signals a shift in industry dynamics. This overview provides a clear breakdown of TurboQuant’s impact on AI accessibility, cost structures and future adoption trends. Key Innovations Behind TurboQuant …

Meta’s new structured prompting technique makes LLMs significantly better at code review — boosting accuracy to 93% in some cases

Meta’s new structured prompting technique makes LLMs significantly better at code review — boosting accuracy to 93% in some cases

Published by skeptic

Deploying AI agents for repository-scale tasks like bug detection, patch verification, and code review requires overcoming significant technical hurdles. One major bottleneck: the need to set up dynamic execution sandboxes for every repository, which are expensive and computationally heavy.  Using large language model (LLM) reasoning instead of executing the code is rising in popularity to bypass this overhead, yet it frequently leads to unsupported guesses and hallucinations.  To improve execution-free reasoning, researchers at Meta introduce “semi-formal reasoning,” a structured prompting technique. This method requires the AI agent to fill out a logical certificate by explicitly stating premises, tracing concrete execution paths, and deriving formal conclusions before providing an answer.  The structured format forces the agent to systematically gather evidence and follow function calls before drawing conclusions. This increases the accuracy of LLMs in coding tasks and significantly reduces errors in fault localization and codebase question-answering.  For developers using LLMs in code review tasks, semi-formal reasoning enables highly reliable, execution-free semantic code analysis while drastically reducing the infrastructure costs of AI coding systems. Agentic code reasoning …

Los Alamos neutron detector boosts accuracy in extreme radiation

Los Alamos neutron detector boosts accuracy in extreme radiation

Published by skeptic

A research team at Los Alamos National Laboratory has unveiled a new neutron detector designed to deliver accurate measurements across a wide range of radiation conditions, addressing long-standing technical and supply challenges in neutron detection. The system, known as the Integrated Composite Optical Neutron Sensor (ICONS), is currently patent-pending and is intended to operate reliably in both low-background environments and high-radiation settings. The development reflects growing demand for tools that can measure neutrons with precision in applications ranging from nuclear security to advanced energy research. Addressing a persistent measurement problem Accurately measuring neutrons has historically been difficult due to the nature of the particles themselves. Unlike charged particles, neutrons do not interact easily with matter, making detection inherently complex. The challenge is compounded by environmental variability: in some scenarios, neutron levels are extremely low, while in others they spike dramatically. Background radiation adds another layer of difficulty. Gamma radiation, which often accompanies neutron emissions, can obscure signals and lead to inaccurate readings in conventional systems. As a result, neutron detection technologies must be both highly …

This Groundbreaking Omega Watch’s Accuracy Is Calibrated Using Sound

This Groundbreaking Omega Watch’s Accuracy Is Calibrated Using Sound

Published by skeptic

Omega has been making the Constellation for more than 70 years, a watch that was the Swiss brand’s elegant flagship timepiece before the Speedmaster landed in 1957. It got its moniker from an image on its caseback—an observatory beneath eight stars. The stars symbolized two chronometer records and six first-place precision awards that Omega earned between 1933 and 1952, the year the Constellation launched. But Omega’s new Constellation Observatory collection has a completely new procedure for measuring accuracy, one that gets around the thorny issue of these pieces having no seconds hand. Why should a seconds hand matter? Watches are tested for accuracy with photographic tracking of the seconds hand over a period of time. Having no seconds hand makes this impossible. The Constellation Observatory pieces, however, grant Omega some watchmaking history as they are the first two-hand watches to achieve Master Chronometer certification without a seconds hand. Traditional testing by COSC—the Swiss body that certifies the accuracy of Swiss watches—uses photographic tech to measure the position of the hands in different positions and temperatures …

iOS 26.4 RC Fixes iPhone Keyboard Accuracy Bug: What’s New

iOS 26.4 RC Fixes iPhone Keyboard Accuracy Bug: What’s New

Published by skeptic

Apple has officially introduced iOS 26.4 RC (Release Candidate) for developers and public beta testers, addressing critical issues that have impacted iPhone users since the release of iOS 26. This update not only resolves a persistent keyboard accuracy bug but also prepares the platform for the launch of the next-generation AirPods Max. Alongside iOS 26.4 RC, Apple has simultaneously updated iPadOS, macOS, tvOS, HomePod software, visionOS, and watchOS, making sure a unified and optimized experience across its ecosystem. Who Can Access iOS 26.4 RC? If you are a developer or public beta tester, iOS 26.4 RC is now available for download. This release represents the final stage of testing before the official public rollout, offering users an early glimpse into the improvements and fixes. With a download size exceeding 8GB, this update performs a complete operating system overwrite, underscoring Apple’s dedication to resolving persistent issues and enhancing performance. By participating in this testing phase, you can provide valuable feedback to help refine the final version, making sure a smoother experience for all users. Key Fix: …

iOS 26.4 Fixes iPhone Keyboard Accuracy Bug

iOS 26.4 Fixes iPhone Keyboard Accuracy Bug

Published by skeptic

The iOS 26.4 update that Apple plans to release as soon as next week includes improvements for the built-in iOS keyboard. In its notes for the software, Apple says iOS 26.4 offers “improved keyboard accuracy when typing quickly.” It’s not entirely clear what Apple means by improved keyboard accuracy, but it’s likely a fix for an iOS keyboard bug that was highlighted on YouTube late last year. When typing some words, the autocorrect keyboard sometimes inexplicably inputs the wrong letter even though the user typed the correct letter, leading to typos. The YouTube video pointing out the issue received over a million views, and it was also further publicized by news sites. There were thousands of comments from people experiencing the problem. It sounds like iOS 26.4 addresses the root issue, preventing the keyboard from inserting the wrong letter when the user is typing quickly. If you’ve experienced issues with the iOS keyboard that have been fixed in iOS 26.4, let us know in the comments below. Popular Stories iOS 27 Will Reportedly Be Like …