11 Oct 2025: AI novel that regenerates daily based on news; Sam Altman on platform strategy; AI and the end of thinking; Chrome's built-in AI model; Big new AI reports

The Next Four Years


Interesing concept: an AI-written novel that regenerates daily based on recent news. As I look at it today on 11 Oct 2025, chapter 1 starts on 15 Oct 2025, with a scientist at the Centres for Disease Control seeing concerning H5N1 bird flu virus mutation rates, in the context of massive budget and job cuts at the CDC. A good counterpart to the 27 Sept post that looked at AI superforecasting.
This experiment set out to answer two questions:
First, can Al analyse eight months of U.S. government upheaval and write a near-term speculative fiction novel that predicts an imaginable future for America?
And next, can Al automatically update that novel daily based on the 24-hour news cycle without any human editorial intervention?
The credits list "Author: Claude Sonnet 4 | Editor: Gemini 2.5 Pro | Researcher: GPT-5". Thanks to Webcurios for recommending this.

An Interview with OpenAI CEO Sam Altman About DevDay and the AI Buildout

In attempting to make sense of the bilzzard of OpenAI's recent launches, this conversation between Ben Thompson (who writes Stratechery) and Sam Altman is useful. Stratechery has been a great source of technology strategy thinking over the years. The latest take compares ChatGPT to Windows' rise to dominance (OpenAI’s Windows Play), with popularity among users attracting developers:

This is a push to make ChatGPT the operating system of the future. Apps won’t be on your phone or in a browser; they’ll be in ChatGPT, and if they aren’t, they simply will not exist for ChatGPT users.

There's lots of interesting quotes in the interview, just picking out a few here. On copyright issues in products like Sora, he's predicting that rights holders will actually want their IP and content to be used:

I predict in another year, maybe less or something like that, the thing will be, “OpenAI is not being fair to me and not putting my content in enough videos and we need better rules about this”, because people want the deep connection with the fans.

And in terms of what's next:
We are going to spend a lot on infrastructure, we are going to make a bet, the company scale bet that this is the right time to do it. Given where we are with the research, with our business, with the product, what we see happening and is it the right decision or not? We will find out, but it is the decision we’re going to make.
Give us a few months and it’ll all make sense and we’ll be able to talk about the whole — we are not as crazy as it seems. There is a plan. ... I do feel like this is a once in a lifetime opportunity for all of us and well take the run at it.


Two links that are worth reading together. The first is a polemical essay by Derek Thompson (journalist and co-author of Abundance with Ezra Klein). He looks at declining writing and reading (in the US) and sees AI as the latest phenomenon after TV, the web, social media, smartphones and then streaming media that "steals our focus" and encroaches on space for deep thinking.
Do not let stories on the rise of “thinking machines” distract you from the real cognitive challenge of our time. It is the decline of thinking people.
The second is a glimpse into China's encouragement of AI in education, come what may: "Beijing is making AI education mandatory in schools", "Guangxi province has instructed schools to experiment with AI teachers, AI career coaches, and AI mental health counsellors". China will be where we first see how Thompson's concerns play out.
The one-foot tall AlphaDog ... was developed by robotics startup Weilan and is powered by DeepSeek’s AI model. In addition to practicing English with Wu’s son, it chats with him about current events, dances to his guitar music, and, through its built-in camera, helps Wu monitor the home when she is away. It has become a part of the family... “My son needs company, but we are a one-child family,” Wu said. “He asks the dog about all kinds of things — national news, weather, geography. Through AlphaDog, he is learning what the world is like.”

How to Try Chrome’s Hidden AI Model

I hadn't realised that Chrome is already shipping with a fully functional local LLM (the tiny but still multimodal Gemini Nano, that also ships with some Android phones). This post explains how to activate and access it. This kind of distribution and usage will be a lot easier for less technical folks compared to using something like Ollama, and will be a disruptive direction if it gets take up (using a local LLM from or within a web page is quite different to installing an app). Very small models that can run on laptops or phones are advancing rapidly but get less press: look at the new 3B parameter Jamba reasoning model from AI21 or the much smaller 7M parameter Tiny Recursion Model from Samsung's AI lab in Montreal.

Important new reports out recently:
  • The annual State of AI from Nathan Benaich and Air Street Capital is always comprehensive and insightful. It is also huge, with a 313-slide deck. Recommended.
  • A report on the State of AI-assisted Software Development from DORA. DORA is the DevOps Research and Assessment group, with a very long running research programme on software development (acquired by Google in 2018). A lot in here about the practicalities and cultural aspects of real life AI adoption in software teams.