Explained
guides, explainers and opinion for the AI-curious
Apple Researchers Built a Tool That Writes Its Own AI Tests
A new system called Agent Seer can automatically generate realistic test scenarios for AI agents by reading the descriptions of the tools those agents use, no human writing required.
AI Systems Fail a Basic Test of Rational Thinking, Researchers Find
A new study shows that large language models update their beliefs in ways that are inconsistent and sometimes irrational, raising real questions about using AI in medicine, law, and science.
Apple Researchers Build a Smarter Marking Scheme for AI Answers
A new training method from Apple ML Research uses detailed, question-specific rubrics to teach AI models how to give better, more trustworthy answers to open-ended questions.
Readers Remind Us: The Terminator Warned About AI Back in 1984
A pair of letters in The Guardian spark a quiet but pointed observation: decades before today's AI anxiety, a cold-war research group and a Hollywood robot both saw this coming.
Nobody Knows What AGI Means, and That's the Point
Nvidia's Jensen Huang declared his company had 'achieved AGI' on an earnings call, then called the milestone 'senseless' in the same breath. He's not wrong. The term has no agreed definition, and tech leaders like it that way.
Apple researchers found a smarter way to build smaller AI models, and it could cut your phone's battery drain
A new technique called IDEA Prune trains a big AI model first, then trims it down. The result is a leaner model that outperforms ones built small from the start.
Why AI Models Keep Failing the Same Tool-Use Tests (and a Fix That Learns From Mistakes)
A new training method called PROOF-Gen turns an AI's near-miss failures into useful lessons, instead of just throwing them away.
A new tool lets you train a medical-grade AI search model on a single gaming GPU in half a day
Sentence Transformers v6.0 adds a smarter way to match search queries to documents, token by token. A real-world test beat every off-the-shelf retrieval model on medical text.
Apple's AI Model Can Write Text and Draw Pictures at the Same Time, Using One Engine
A new research model from Apple blends language and image generation into a single, unified system. Here is why that matters for the next generation of AI assistants.
Will AI replace mathematicians? Probably not yet, say experts
About 40 top mathematicians met at OpenAI's offices this month to talk about their profession's future. The mood was worried. The evidence, though, tells a more complicated story.
Gradio Launches Visual AI Pipeline Builder That Doubles as a Live API
A new tool called gr.Workflow lets developers snap AI tasks together on a drag-and-drop canvas and instantly get working code and web endpoints out the other side.
L'IA impara a pensare in immagini internamente, riducendo i tempi di analisi video senza perdere accuratezza
Un nuovo metodo di addestramento chiamato Internalized Visual Thinking consente ai modelli di IA di ragionare sui video come gli umani visualizzano le cose nella loro mente, saltando il lento passaggio di generare immagini effettive durante l'analisi.
AI Is Reshaping Engineering Jobs. Here Is What Actually Keeps You Employable
Experts say the secret is not learning the right coding language. It is learning how to keep learning, whatever comes next.
I bambini imparano le lingue con una frazione dei dati di cui l'IA ha bisogno. Gli scienziati vogliono capire perché.
Un bambino piccolo padroneggia la grammatica dopo aver sentito circa 10 milioni di parole. Un grande modello linguistico può consumarne 15 trilioni. Colmare questo divario potrebbe trasformare il modo in cui l'IA viene costruita.
This Tech Conference Is Worth Binge-Watching Like Your Favourite TV Series
Software Should Work 2026 packs 13 talks on code reliability into a single watchable day. It might be the best thing on your screen this summer.