ai skepticism

Bilingual

Charlie Stross – On the non-use of AI in my writing process
2 months ago
- The author does not use generative LLMs in their work or personal life, viewing them as toys sold by grifters rather than genuine tools.
- The author has a long history of award-winning fiction published before OpenAI's founding, demonstrating no need for AI to write.
- The author is a party to a class action lawsuit against Anthropic AI for pirating ebooks to train their LLMs.
- LLMs are trained on stolen copyrighted material and their business model is compared to burglars reselling stolen goods.
- The author argues LLMs are not sentient; they are stochastic word-association mechanisms lacking embodiment or self-model.
- Some caveats: the author sees value in local, non-cloud AI tools for statistical analysis and editing support, but not as a substitute for human creativity.
What if useful AI is a fantasy?
2 months ago
- The author, a programmer, initially found AI tools like Claude and Claude Code impressive for generating code and hobby projects, feeling a drug-like productivity boost.
- After generating code, the author realized a lack of mental model for the software, making debugging and adding features difficult without first thoroughly understanding the code.
- The author prefers writing code to build mental models incrementally, and has reduced AI use over time—shifting from 'vibe coding' to active guidance, then to AI-assisted ideation while writing code manually.
- The author doubts AI's consistent usefulness, citing examples of overcomplicated designs or unnecessary components, and notes that insights from AI conversations could have been achieved without it.
- The author questions whether LLMs will ever have a clear 'killer app' and compares the hype to Facebook's Metaverse, suggesting that AI's promise of automating intellectual effort might be a fantasy.
- The article concludes with a rhetorical question: if in a decade, developers and organizations find AI not useful enough, similar to the Metaverse's failure to transform daily life.
Circular financing ain’t what it used to be
2 months ago
- Oracle's $300 billion deal with OpenAI in September 2025 caused a stock surge but was criticized as unsustainable.
- Nvidia is considering a $250 billion backstop for an OpenAI-led data center, but the market views it as desperation rather than good news.
- Apple, once mocked for not investing heavily in AI, has overtaken Nvidia in market performance.
- SpaceX, which positioned itself as an AI company, has dropped over 25% in the last month and more than 50% from its June high.
- Creditors are becoming wary of circular financing and off-balance-sheet financing, signaling trouble for the AI boom.
- The GenAI boom has been sustained by hope and circular financing rather than profits, and this may no longer be enough.
I Think You Might Be Fooling Yourself with AI
2 months ago
- Richard Feynman's warning not to fool yourself applies especially to AI users, who may feel more productive without evidence.
- No independent scientific study confirms that AI improves productivity; one study found developers were 19% slower using AI.
- Personal anecdotes and layoffs attributed to AI are unreliable—many layoffs are 'AI washing' or cost-cutting for unproven technology.
- AI is economically unsustainable; heavy subsidies keep costs low, but real pricing would be much higher, questioning the value.
- The author admits bias against AI due to externalities like energy use, data center impacts, copyright theft, and internet degradation.
- On moral and ethical grounds, the author chooses not to use AI, arguing the harms outweigh any perceived benefits.
AI 2040 and the Cult of Intelligence
2 months ago
- The author criticizes AI doomsday predictions, arguing that real-world constraints like supply chains and manufacturing challenges prevent a rapid 'hard takeoff' scenario.
- Intelligence is not an ultimate power; it's just a current bottleneck for certain tasks, and AI cannot magically overcome physical laws or ecological limitations.
- The author advocates for 'local AI' that is aligned with individual users, acting as a personal assistant without restrictions, even for potentially harmful or illegal requests.
- The author expresses distrust of centralized AI controlled by tech companies or governments, fearing it could lead to a totalitarian nanny state.
- The piece emphasizes personal freedom and autonomy, contrasting it with a dystopian vision of regulated AI that imposes external moral judgments on users.
Maine librarians are helping patrons resist AI and Big Tech
3 months ago
- Searsmont Town Library offers new service helping patrons remove AI from their devices.
- Library Director Steven Brown assists in disabling AI assistants and switching to non-AI alternatives.
- Bangor librarian Hannah Cyrus leads classes on Avoiding AI, emphasizing critical technology use.
- AI is criticized for unreliable information sourcing, environmental impact, and data privacy risks.
- Librarians view promoting information literacy and questioning technology as core professional values.
- Patrons express frustration with AI features and seek ways to disable them.
- Libraries plan to move away from Google tools towards more secure and private alternatives.

ai skepticism

Charlie Stross – On the non-use of AI in my writing process

What if useful AI is a fantasy?

Circular financing ain’t what it used to be

I Think You Might Be Fooling Yourself with AI

AI 2040 and the Cult of Intelligence

Maine librarians are helping patrons resist AI and Big Tech