Ad
Short

Eighty-one-year-old psychologist Harvey Lieberman describes ChatGPT as "not a crutch, but a cognitive prosthesis — an active extension of my thinking process."

In a recent New York Times essay, Lieberman explains how an experiment with ChatGPT turned into a daily routine. He treats the AI as a reliable thinking partner, using it to sharpen his language, deepen self-reflection, and even spark emotional resonance. At a stage in life when thoughts can slow down, Lieberman says ChatGPT has helped him "re-encounter my own voice."

"ChatGPT may not understand, but it made understanding possible."

Dr. Harvey Lieberman

ChatGPT has also drawn criticism for reinforcing users' beliefs and, in some cases, steering vulnerable or mentally ill people into negative thought patterns. OpenAI has acknowledged these risks.

Short

Cohere's new Command A Vision model is designed to handle images, diagrams, PDFs, and other types of visual data. Cohere says the model outperforms GPT-4.1, Llama 4 Maverick, Pixtral Large, and Mistral Medium 3 on standard vision benchmarks.

The model's OCR can recognize both the text and the structure of documents such as invoices and forms, outputting the extracted data in structured JSON. Command A Vision can also process real-world images, like identifying potential risks in industrial environments, the company says.

Image: Cohere

Command A Vision is available through the Cohere platform and for research on Hugging Face. The model can run locally with either two A100 GPUs or a single H100 using 4-bit quantization.

Ad
Ad
Ad
Ad
Short

Google is expanding its "AI Mode" to the UK, following launches in the US and India. The new feature appears as an extra tab in Google Search and in the Google app for Android and iOS, letting users ask complex questions via text, voice, or image and get AI-generated answers with additional links. AI Mode runs on a customized version of Gemini 2.5 and uses query fan-out techniques to break down questions into smaller subtopics, searching them all at once.

Google says AI Mode should result in a "greater diversity" of visited websites—a phrase that sidesteps the more important effect: less traffic flowing to the open web. Even the "light" version, called AI Overviews, has already led to a sharp drop in web clicks.

Short

Starting August 28, 2025, Anthropic will roll out new weekly usage limits for Claude subscribers. Alongside the existing 5-hour limit - which resets every five hours - users will face two additional restrictions: a general weekly cap for all models and a separate weekly limit for Claude Opus 4. Both reset every seven days.

For Claude Sonnet 4, Anthropic estimates weekly usage will fall between 40 and 80 hours, depending on project size and settings like auto-accept mode. Users running multiple Claude Code instances at once will hit the new limits sooner.

Anthropic says fewer than five percent of users will be affected. The company cites abuse prevention - including account sharing and continuous use - as well as overall system performance as reasons for the change.

Short

Microsoft is rolling out a new Copilot mode for its Edge browser. Mustafa Suleyman describes it as a streamlined, AI-powered interface with voice control that can automatically recognize URLs, search terms, and chat commands. Copilot analyzes the context of multiple open tabs, making it easier to compare travel plans or find the right restaurant without switching between windows.

Video: Microsoft

Copilot now sits at the top left of the Edge browser and is available for free. Microsoft says upcoming features will include project-based history sorting and contextual recommendations.

Ad
Ad
Google News