Saltar al contenido

Artículos y prensa

Apariciones en medios, artículos publicados y reflexiones sobre gestión de proyectos, IA y tecnología.

Prensa y apariciones

Artículos en Medium

Fable 5, the Cost of Competing, and the Suspicion That Something Changed

26 de agosto de 2026

Fable 5, the Cost of Competing, and the Suspicion That Something Changed

Claude Fable 5 has run under three different access designs since it launched, with a global blackout over export controls in the middle:

Leer en Medium →

Uncensored AI Models Are a Real, Present Danger

25 de agosto de 2026

Uncensored AI Models Are a Real, Present Danger

Right now, anyone with a 24 GB graphics card can download a genuinely capable AI model with its safety layer stripped out and run it offline. This week, developer Alex Finn tested exactly that: he ask…

Leer en Medium →

Europe Is Raising the Cost of Keeping Data-Centre Options Open

21 de agosto de 2026

Europe Is Raising the Cost of Keeping Data-Centre Options Open

Data-centre development has always depended on more than electricity prices. Land, fibre, planning, taxation, water, grid capacity and local politics all matter, and serious developers have never sele…

Leer en Medium →

Peer Review Is Broken. The Owners Can Afford to Fix It.

20 de agosto de 2026

Peer Review Is Broken. The Owners Can Afford to Fix It.

Six months after I submitted a paper to a well-established, Q1-ranked journal, the editor told me to withdraw it: not because the methodology was weak, the fit was poor, or the idea lacked novelty, bu…

Leer en Medium →

Claude’s Watermark: What the Removal Tools Are Actually Doing

19 de agosto de 2026

Claude’s Watermark: What the Removal Tools Are Actually Doing

Claude can now generate text carrying an invisible watermark, part of Anthropic’s response to the EU AI Act’s transparency rules. That much has been covered widely. The more useful question now is how…

Leer en Medium →

AI Providers Encrypted Their Models’ Reasoning. It Leaked Anyway

18 de agosto de 2026

AI Providers Encrypted Their Models’ Reasoning. It Leaked Anyway

Reasoning models can generate hidden chains of thought before producing the answer shown to the user, and what shows up in the chat window is usually not the whole of that process. Providers increasin…

Leer en Medium →

What PMI’s 72% Framework Success Rate Really Measures

17 de agosto de 2026

What PMI’s 72% Framework Success Rate Really Measures

PMI’s Pulse of the Profession 2026, titled “Driving Success in Complex Projects: From Navigating Tasks to Navigating Systems,” reports that project professionals who apply a structured complexity fram…

Leer en Medium →

Five Months After the SaaSpocalypse, Software Split Into Winners and Losers

27 de julio de 2026

Five Months After the SaaSpocalypse, Software Split Into Winners and Losers

In February, I wrote about the week ServiceNow and Microsoft both beat earnings estimates and their stocks fell anyway. The same week, the iShares Expanded Tech-Software ETF entered a bear market, its…

Leer en Medium →

Opus 5 Does Not Stop at the First Plausible Answer

25 de julio de 2026

Opus 5 Does Not Stop at the First Plausible Answer

Anthropic gave Claude Opus 5 a drawing of a machine part and asked it to write code that would rebuild the part as a 3D model in FreeCAD. The test deliberately withheld any way for the model to view t…

Leer en Medium →

Estos son los 9 más recientes. Hay más en Medium.

Ver todos los artículos en Medium →