ai

The “magic” might be remarkable, Alberto Romero writes, but it shifts the effort to the wish. And most people (and organizations) aren’t very good at that part.

What do you do if you find the whole AI enterprise so immoral that you find its very use to be ethically unacceptable?

A BCG study of US-based workers found that intensive oversight of AI agents can cause cognitive exhaustion that the researchers call “AI brain fry.”

Ethan Mollick reviews the research that chat interfaces introduce heavy effort that undermines complex, specialized work. He says the future is radically adaptive interfaces.

Nicholas Evans on LinkedIn: What happens when we go all-in on a single AI interface to do all the things.

Fin researcher Genna Bridgeman shares practical findings about how AI interactions are affected by social expectations of the specific communication channel.

Benedict Evans deflates the frothy talk that AI agents and assistants will eliminate vast swaths of software.

Ethan Mollick notes that AI has begun to unlock radical changes in the shape of work. What that shape will ultimately become is uncertain—which means now is the moment to influence it.

The banking app launched a fun and smart campaign to educate about the risks of deepfake scams.

TJ Pitre explains how to bake context into your design system, which pays dividends in AI-powered automation but also next-generation Sentient Design experiences.

Scott Jenson suggests AI is likely to be more useful for “boring” tasks than for fancy outboard brains that do all our thinking for us.

Sharif Shameem pulled together a wishlist of fun ideas for AI-powered applications.

TechCrunch gathered a collection of apps that are using Apple’s on-device models to provide casual intelligence that’s free, easy, and priveate to the user.

Anton Sten: While AI hardly created the problem of ill-considered products, it will certainly accelerate them.

A research paper finds that the reliability of large language models plummet the longer a conversation continues.

Toby Ord finds that an AI agent’s chance of success drops off exponentially the longer a task takes.

The Wall Street Journal: “Websites in general will evolve to serve primarily as data sources for bots that feed LLMs.”