Prepare - Refine - Scale (with some caveats)

OpenAI and Anthropic spend billions of dollars a year training models like GPT-4 and Claude, but competitive price dumping is making the business around

I've been playing around with OpenAI's Advanced Voice Mode for the last week, and it's the most convincing taste I've had of an AI-powered future yet. OpenAI's newest feature, currently in a limited alpha test, doesn't make ChatGPT any smarter than it was before. Instead, Advanced Voice Mode (AVM) makes it friendlier and more natural to talk to. It creates a new interface for using AI, and your devices, that feels fresh and exciting; and that's exactly what scares me about it.

Anthropic has published the system prompts for Claude, its family of generative AI models. They instruct the model how it should -- and shouldn't -- behave.

Researchers have long disagreed over what constitutes open-source AI. An influential group has offered up an answer.

Victor Miller proposed customized ChatGPT bot to govern Cheyenne, Wyoming – but fared badly at the ballot box

Companies and organisations need a clear etiquette for online meetings, when it comes to the...

Nature - The rising energy and environmental cost of the artificial-intelligence boom is fuelling concern. Green policy mechanisms that already exist offer a path towards a solution.

Sakana AI's groundbreaking 'AI Scientist' system autonomously conducts end-to-end scientific research, challenging traditional scientific methods and raising questions about the future of discovery.

SAG-AFTRA has struck a deal for voice replicas with AI firm Narrativ that the union asserts sets 'a new standard' for ethical use of the technology.

LLMs are known for reasoning powers. But new research from MIT and UCLA shows LLMs differ widely between inductive and deductive reasoning.

An MIT team used probing classifiers to investigate if language models trained only on next-token prediction can capture the underlying meaning of programming languages. They found that it forms a representation of program semantics to generate correct instructions.

Deep in the heartlands of Brazil, Indigenous creators are having to censor themselves to avoid getting banned on social platforms.

The proliferation of friendly, often sexually suggestive, chatbots is ‘making it so easy to make a bad choice,’ one observer says.

This is what an old lesbian looks like: the huge, joyful project capturing queer elder women’s stories before they’re lost

Researchers at MIT have released the AI Risk Repository, a comprehensive database that can help organizations identify and mitigate AI risks.

The effectiveness of teamwork in professional and academic environments typically depends on various factors, including communication and coordinated attention. In this context, collective attention entails ...

Processing these simultaneous truths — inspiration on the one hand, and systemic injustice on the other — feels like a microcosm for our wider dilemma. As we co-create a new/old system, a pluriverse of alternatives, what do we want to keep — and what will we leave behind? What might a post-growth Olympic Games look like?

Critics say a proposed change could allow large energy users to hide their true emissions

Andrew Hick - accessibility and test person. User of code, pixels, pens and string. Bath, UK

Manda Scott's original new novel, 'Any Human Power' is an exploration of how we might transition to another world that is possible.

Anthropic is opposing Scott Wiener’s California bill to regulate AI, taking Google and Amazon money, and scraping web data without permission.

More tasks, human baselines, and preliminary results for GPT-4 and Claude.

How the artificial intelligence industry profits from an unprotected digital working class in Brazil Workers perform crucial tasks for the development of AI systems...

Climate Spring presents its new climate storytelling guide which sets out what makes a good climate story, by shifting narratives from apocalypse to regeneration.

Design guidelines following good practices that you can apply to product design, from components to design systems.

The Results Part 3: Creating a Course Outline

Perhaps we should be promoting AI-resilience instead