Browsing: LLMs

Apple is planning to launch an artificial intelligence-powered upgrade of Siri in spring 2026, after facing delays and technical challenges

Apple struggles to upgrade Siri with cutting-edge AI technology

The AI community is increasingly using ‘vibes’ to assess generative AI and large language models, rather than conventional quantitative metrics. This trend is raising concerns about the objectivity and reliability of AI evaluations.

A recent study from the University of Edinburgh highlights a significant gap in AI’s ability to perform simple time-related tasks, such as reading analog clocks and understanding calendars.

Anthropic CEO Dario Amodei suggests giving AI models the ability to quit jobs they find undesirable, raising questions about anthropomorphism and the true nature of AI.

This article examines the potential for generative AI and large language models to detect and report on “thought crimes” based on user dialogue, exploring the ethical implications, technological capabilities, and societal impact of such a scenario. It highlights potential downsides, including false positives and the chilling effect on free speech.

Discover a new method for training AI to use logical reasoning effectively, using expert analysis of AI and LLMs. Learn how generative AI can improve its ability to think logically, and the importance of sound reasoning.

An examination of recent claims that generative AI and large language models (LLMs) suffer from cognitive decline, exploring the nuances of AI development and comparing it to human cognitive processes.

Small Language Models (SLMs) are emerging as a powerful alternative to large language models (LLMs), offering advantages in environmental impact, cost-effectiveness, accessibility, and privacy. This article explores the benefits of SLMs and why ‘assistive intelligence’ is a more apt description.

Microsoft researchers introduce OmniParser V2, a significant upgrade to the original tool, designed to enhance GUI automation by improving accuracy in identifying interactable elements and speeding up inference.