Author: Advanced AI Editor

Diagnosis and management of different ophthalmic subspecialties were better addressed when clinical cases were entered into DeepSeek-R1, which outperformed OpenAI o1 in the same categories. The study, published in JAMA Ophthalmology,1 also found that DeepSeek-R1 was able to lower operating costs with its performance in diagnosis and management.DeepSeek outperformed OpenAI in clinical diagnosis and management of ophthalmic subspecialties | Image credit: 光画社 (Kōgasha) – stock.adobe.comClinicals can be aided with clinical decision-making with the use of large language models,2 a form of AI that has natural language processing capabilities. DeepSeek-R1 is a form of large language model that was released in…

OpenAI today confirmed that it has signed an agreement to purchase computing infrastructure worth $300 billion from Oracle Corp. over a period of roughly five years, as part of its previously announced Project Stargate data center building initiative. The deal represents one of the biggest cloud contracts in history. It’s a massive commitment that far outstrips OpenAI’s current revenue, reflecting the frenzied rate of spending on artificial intelligence data centers by Silicon Valley companies. OpenAI and Oracle will construct 4.5 gigawatts of data center compute capacity, which is roughly equivalent to the electricity produced by more than two Hoover Dams, or the…

International Business Machines Corporation IBM and Qualcomm Incorporated QCOM are tech legacy firms pivoting toward cutting-edge growth areas of either cloud computing and AI (artificial intelligence) workloads, or chips and patent-driven innovations. IBM offers cloud and data solutions that aid enterprises in digital transformation. In addition to hybrid cloud services, the company provides advanced information technology solutions, computer systems, quantum computing and supercomputing solutions, enterprise software, storage systems and microelectronics.Qualcomm offers high-performance, low-power chip designs for mobile devices, PCs, XR (Extended Reality), automotive, wearable, robotics, connectivity and AI use cases. The company boasts a comprehensive intellectual property portfolio comprising 4G,…

Anthropic reported a service outage impacting APIs, Console, and Claude earlier this afternoon.  Users on GitHub and Hacker News noted issues with Claude at around 12:20 p.m. ET, with Anthropic releasing a status update eight minutes later, noting that its APIs, Console, and Claude AI were down. At press time, the company said it had implemented several fixes and was monitoring the results. “We’re aware of a very brief outage of our API today shortly before 9:30 a.m. PT,” an Anthropic spokesperson told TechCrunch. “Service was quickly restored.” Anthropic is no stranger to errors or bugs on its platform and…

NEWSLETTER INSIGHTS 11 September, 2025 | by Patrick Overall 73% of customers say experience is a deciding factor in their purchasing decisions, right behind price and product quality. They’re willing to pay more for the qualities that matter most: greater convenience, friendly and welcoming interactions and moments that leave a lasting impression. That’s why 38% of business leaders now cite improving customer experience as the primary focus of their generative AI investments. AI is becoming a powerful tool to help frontline teams deliver faster, more personalized and more memorable service without losing the human connection that keeps customers coming back.…

By Sabrina Pervez, SpotDraft. We all know by now: the EU AI Act is live, fines are eye-watering, and AI vendors are under the microscope. But here’s the twist: if you’re a General Counsel or Chief Legal Officer, the risks are multifaceted: lost business opportunities, scrutiny over how your data is processed, damage to your reputation, and, ultimately, the possibility of your vendor facing a hefty fine. The smartest legal leaders aren’t waiting until 2026. They’re already grilling vendors on compliance and future-proofing. Why? Because the EU AI Act is changing enterprise procurement today, not two years from now. Law…

Stephen Ehikian, the acting head of the General Services Administration for the first six months of Trump 2.0 until his recent departure, says he’s left the agency in a better position than when he started. He exited GSA a week ago to join the private sector and lead C3 AI, after being replaced by a State Department official, Michael Rigas, in late July. Shortly after Ehikian was sidelined, President Donald Trump nominated financial executive Ed Forst to officially lead the agency, pending Senate confirmation. Some have interpreted the changes as a reining in of Department of Government Efficiency-affiliates, including Ehikian and GSA’s Federal Acquisition…

YouTube announced on Wednesday that its multi-language audio feature has officially launched after a two-year-long pilot. Now, millions of YouTubers can add dubbing to their videos in different languages, helping them reach a wider global audience. The rollout is expected to happen over the coming weeks. The feature initially launched as a pilot in 2023, available to a limited number of creators, including MrBeast, Mark Rober, and chef Jamie Oliver. Creators had to work with third-party dubbing services until YouTube introduced an AI-powered auto-dubbing tool that leverages Google’s Gemini technology to replicate a creator’s tone and emotions. Since its launch,…

Arbitration-focused platform Jus Mundi has launched what it’s calling Jus AI 2, an AI-powered legal research capability that combines agentic reasoning with deterministic search control. The company kindly explained to Artificial Lawyer how its agent properties work, (see below). The move builds on an earlier AI product phase, which they’ve called Jus AI 1. The new solution’s ‘AI planning agent creates a multi-step research plan tailored to each query. It then analyses up to 75,000 documents per minute, identifying the most relevant legislations, precedents, and publications from all over the world’. And ‘it delivers a structured, actionable answer, with transparent…

Parallel thinking has emerged as a novel approach for enhancing the reasoning capabilities of large language models (LLMs) by exploring multiple reasoning paths concurrently. However, activating such capabilities through training remains challenging, as existing methods predominantly rely on supervised fine-tuning (SFT) over synthetic data, which encourages teacher-forced imitation rather than exploration and generalization. Different from them, we propose \textbf{Parallel-R1}, the first reinforcement learning (RL) framework that enables parallel thinking behaviors for complex real-world reasoning tasks. Our framework employs a progressive curriculum that explicitly addresses the cold-start problem in training parallel thinking with RL. We first use SFT on prompt-generated trajectories…