Jalapeño is not just the first chip designed by OpenAI. It is a strong signal that artificial intelligence is entering the design cycle of its own infrastructure, potentially giving OpenAI an advantage that could become extremely difficult for others to close. When OpenAI introduced Jalapeño last June, the most immediate interpretation was fairly predictable: OpenAI … Read more
AI could reduce the number of developers needed inside large organizations while, at the same time, creating new demand for technical skills in companies that until now have never been able to afford a real software department. When we talk about the impact of artificial intelligence on developers’ jobs, the discussion almost always tends to … Read more
A few days ago, Unsloth made Qwen3.8-27B available, and I downloaded it to run locally with Ollama. I had heard about the benchmarks for this model, so naturally it caught my attention right away. Not because running an LLM on your own computer is still something new, but because we are starting to see relatively … Read more
For most of my professional life, the advice I received was fairly simple: specialize. The mantra was always something along the lines of: “Find something you’re good at, go deeper into it than everyone else, and build your value around it”. The more complex the market became, the more logical it seemed to divide work … Read more
When a Model Can Use Tools, Credentials and Networks, a Wrong Answer No Longer Stays on the Screen For a long time, when discussing the risks of artificial intelligence, the most immediate example was hallucination. A model could invent a source, misinterpret a request, or provide an answer that sounded convincing but was completely wrong. … Read more
A conversation about AGI started with the future of software and ended up touching on universal basic income, social mobility, taxation, ownership and power. We did not find the solution, but perhaps we found better questions. This is not an interview in which I asked artificial intelligence to predict the future, nor an attempt to … Read more
In seven days, Codex processed 65 billion tokens for me. That does not mean it produced 65 billion tokens’ worth of value, but it does say something about the way agents are changing both the duration and the intensity of work. A few weeks ago, I installed Who Burned More, a small open-source tool that … Read more
Watermark e Content Credentials possono aiutarci a capire come è stato prodotto un contenuto, ma non possono certificare che ciò che stiamo vedendo o leggendo sia vero e degno di fiducia. Quando ho letto che Anthropic avrebbe cominciato a inserire una marcatura invisibile nei testi generati dai nuovi modelli Claude e metadati di provenienza firmati … Read more
Quando usi un coding agent, la tentazione più naturale è partire dal risultato. “Costruiscimi un’applicazione che faccia questo.“ Descrivi la funzionalità, magari indichi linguaggio e framework preferiti, lasci che l’AI cominci a lavorare. Dopo poco compaiono la struttura del progetto, il database, alcune API, l’interfaccia, forse anche una prima serie di test. Il risultato può … Read more
Quando oggi si parla di sviluppare un’applicazione basata sull’intelligenza artificiale, quasi inevitabilmente la discussione comincia dal modello da utilizzare, dalle funzionalità che dovrà avere l’applicazione, dall’interfaccia e, naturalmente, dal prompt. È comprensibile. Sono le parti immediatamente visibili del prodotto e sono anche quelle che permettono di ottenere velocemente qualcosa da mostrare. Si collega un LLM, … Read more