Today's Hallucination HQGen Z Uses AI Constantly, Feels Terrible About It
A Gallup survey of nearly 1,600 Americans aged 14–29 finds Gen Z is growing increasingly sceptical of AI — its trustworthiness, its effect on learning, its general vibe — while simultaneously refusing to stop using it. This is, it turns out, exactly how most people feel about social media, fast food, and late-night doom-scrolling. The disillusionment is real; the usage figures suggest it is also entirely beside the point.
Source: The Verge
Swiss Researchers Build a Quantum Computer That Actually Behaves Itself
ETH Zurich has demonstrated a 17,000-qubit array of neutral atoms operating at 99.91% fidelity — meaning it gets things wrong roughly once in every ten thousand operations, which puts it well ahead of most software in daily use. The trick is geometric-phase quantum gates, which exploit the shape of a computational path rather than its precise timing, making them unusually resistant to the experimental noise that normally sends quantum systems into a sulk. Quietly extraordinary work.
Source: ETH Zurich
Amazon Is Spending $200 Billion on AI. It Is Currently Making $15 Billion From AI.
In his annual shareholder letter, Amazon CEO Andy Jassy confirmed the company plans to invest roughly $200 billion in AI infrastructure in 2026, against AI revenue currently running at $15 billion annually. The gap between those two numbers is, by any measure, a bold statement of faith. Also mentioned: Project Leo, Amazon's satellite internet play, already has 200+ birds in orbit. AWS, satellites, and a cheery tolerance for very large deficits — it's quite the shareholder letter.
Source: Next Big Future
Linus Torvalds Permits AI Coding Assistants, With The Enthusiasm Of A Man Permitting Pigeons
The Linux kernel — the software backbone of most of the internet — has quietly added official guidance on using AI coding assistants. The document doesn't ban them. It does, however, spend considerable effort explaining how AI-generated code tends to look correct while being subtly wrong, and how contributors remain fully responsible for every line regardless of how it arrived. Permission granted, then. Warmly worded, it is not.
Source: Hacker News / GitHub
Canada and Germany's AI Champions Reportedly Considering a Merger, Berlin Approves
Cohere, the Canadian enterprise AI firm, and Aleph Alpha, Germany's answer to the question "but what if AI, but European," are in merger talks according to Handelsblatt, which cites government and industry sources. Berlin is said to be supportive — unsurprising, given that Europe has been trying to build an AI champion capable of competing with American giants for roughly as long as people have been predicting it would. Whether two mid-sized contenders combine into something formidable, or simply a larger mid-sized contender, remains to be seen.
Source: CNA
As always, the AI was confident, the humans were sceptical, and everyone carried on regardless.
|