AI Labs Shredding Millions of Rare Books to Feed Models
6 Meldungenca. 4 Min. LesezeitArchiv · ungeprüft
Archiv-Ausgabe vor Einführung der QuellenprüfungDiese Ausgabe ist vor dem 17. Oktober 2026 entstanden und wurde nicht nachträglich geprüft. Sie kann Fehler oder überholte Angaben enthalten. Für Entscheidungen bitte die verlinkten Originalquellen lesen.
In einer Minute im Bild
AI labs are buying, scanning, and shredding millions of rare, out-of-print books to train models, destroying irreplaceable cultural artifacts in the process.
Google's Chief Scientist is leaving after 27 years to start a new AI company, signaling a major brain drain from big tech.
A single Wikipedia page costs an AI agent 68,000 tokens, exposing the hidden computational expense of web browsing agents.
AMD and Cerebras launched a new AI inference solution, while Mozilla published its first 'State of Open Source AI' report.
Meldung 1
AI Labs Shredding Millions of Rare Books to Feed Models
AI labs are in an arms race to buy up rare, out-of-print books, slicing them open, scanning them, and pulping the remains. This industrial-scale destruction is erasing the last physical copies of countless texts.
Florians EinordnungThis is a cultural crime dressed up as progress. You don't destroy the only copy of a book to read it—you scan it and put it in a library.
Für deinen Alltag Contact your local library or the Internet Archive to support legislation requiring mandatory deposit of scanned books before physical destruction.
Ganzen Artikel lesen
What can be done? Immediate action is needed from lawmakers to require that any digitized book be deposited in a public archive before the physical copy is destroyed. The cultural loss is irreversible; the data gain is incremental. We are trading our libraries for a slightly better chatbot.
Meldung 2
Google Chief Scientist Exits After 27 Years to Launch AI Startup
A top-tier Google scientist is leaving the company after nearly three decades to found a new AI venture. This marks one of the most significant departures from the tech giant in recent years.
Florians EinordnungWhen the people who built the cathedral leave to build their own chapel, you know the money is moving. Google's talent drain is becoming a flood.
Für deinen Alltag If you are hiring in AI, look for senior researchers who are 'under-utilized' at Big Tech—they are likely planning their exit strategy right now.
Ganzen Artikel lesen
The new company is expected to focus on 'AI safety' or 'agentic AI,' the two hottest buzzwords in the sector. While the details are scarce, investors are already lining up. This move validates the thesis that the most ambitious AI work is no longer happening inside Big Tech, but in nimble startups with access to massive compute budgets.
Meldung 3
One Wikipedia Page Costs Your AI Agent 68,000 Tokens
Researchers calculated the token cost for an AI agent to process a single Wikipedia page, revealing a massive computational overhead. This number highlights the inefficiency of current agent architectures.
Florians EinordnungThis is the dirty secret of the agent economy: they are burning money just to read a paragraph. The infrastructure is not ready for scale.
Für deinen Alltag When building an AI agent, cap the number of 'tool calls' per task to under 10 to avoid token blowout costs.
Ganzen Artikel lesen
This also explains why AI labs are so desperate for clean data—they cannot afford to have agents browsing the messy web. They would rather buy and shred books than pay for a trillion tokens of agentic browsing. The economics of data acquisition are driving the destruction of libraries.
Meldung 4
AMD and Cerebras Team Up to Challenge Nvidia's AI Dominance
AMD and Cerebras announced a partnership to deliver advanced AI inference solutions, combining AMD's software stack with Cerebras's massive wafer-scale hardware. This is a direct challenge to Nvidia's near-monopoly.
Florians EinordnungFinally, a real challenger to Nvidia. Cerebras's hardware is absurdly fast; AMD's software is getting decent. This might actually work.
Für deinen Alltag If you are evaluating AI infrastructure, keep an eye on Cerebras benchmarks for inference—they are likely to be the fastest on the market within six months.
Ganzen Artikel lesen
The real test will be whether they can challenge the CUDA ecosystem. Nvidia's software lock-in is the moat, and breaking it requires more than just a marketing partnership. However, with the pressure from regulators and the demand for alternatives, this is the first credible threat to Nvidia's dominance in over a decade.
Meldung 5
Mozilla's First 'State of Open Source AI' Report Is Here
Mozilla released its inaugural report on the state of open source AI, detailing trends, challenges, and the growing divide between open and closed models. The report highlights the fragility of the open source ecosystem.
Florians EinordnungMozilla is the conscience of the internet, and this report is a wake-up call. Open source AI is dying a death by a thousand cuts—legal, financial, and computational.
Für deinen Alltag Review your company's AI dependencies: if you rely solely on closed APIs, you are exposed to price hikes and policy changes; start planning for open-source fallbacks now.
Ganzen Artikel lesen
For developers, the takeaway is to support organizations like EleutherAI and the Linux Foundation that are working on truly open alternatives. The future of the web depends on having AI models that are not controlled by corporate gatekeepers.
Meldung 6
Andrew Ng Launches LearnVector for Personalized AI Education
Andrew Ng, a prominent AI educator, has launched a new company called LearnVector focused on one-to-one learning experiences powered by AI. The startup aims to replace static courses with dynamic tutors.
Florians EinordnungNg is the best marketer in AI, and he knows the 'course' model is dead. If anyone can make AI tutoring mainstream, it's him.
Für deinen Alltag If you are building an educational product, focus on interactivity and feedback loops—passive content consumption is not a defensible business.
Ganzen Artikel lesen
For learners, this is the beginning of the end of the 'one-size-fits-all' curriculum. Expect to see AI tutors that know your weaknesses and drill you on them, rather than forcing you to watch a 10-hour video series. The quality of education could finally improve, not just the accessibility.
Prompt des Tages
Rare Book Ethics
You are an AI policy researcher. Draft a 3-paragraph memo to a major AI lab's leadership team outlining ethical guidelines for acquiring and digitizing rare books. Include a mandatory clause about preserving the physical artifact after scanning.
Use this to frame internal policy discussions when your team is considering mass digitization of physical media.
Deep Dive
How to Calculate the Real Cost of AI Agents (Before You Build One)
Remember, the 68,000-token figure is a warning, not a benchmark. You can do better, but only if you are deliberate about the design. The future of agents belongs to those who can make them cheap, not just smart.
Don't let the machines eat the books.
Für Ihren Betrieb
Welcher Ablauf lohnt sich bei Ihnen zuerst?
Mit dem kostenlosen KI-Prozess-Check sehen Sie in 15 Minuten, wo in Ihrem Betrieb die meiste Zeit liegen bleibt. Oder wir sehen es uns in 30 Minuten gemeinsam an, kostenlos und ohne Verkaufsgespräch.
