A Co-Created Manifesto for a Future with AI We are delighted to announce that we will be partnering with IDEO
Join us in creating something that doesn't yet exist: a genuinely democratic vision for AI's role in human flourishing.
AI offers speed, but real expertise takes time. Discover how to stay essential by preserving the deep skills that AI can’t replace—and why your value depends on it.
A review of Christopher Summerfield's These Strange New Minds: How AI Learned to Talk and What It Means
These Strange New Minds: How AI Learned to Talk and What It Means
by Christopher Summerfield
Christopher Summerfield’s These Strange New Minds is a must-read for anyone who enjoys the crossovers between artificial intelligence and human intelligence. He's rare in his ability to speak fluently across neuroscience, AI, and philosophy without slipping into abstraction.
What I found most compelling about the book is his perspective on the science behind LLMs—and the insight that language itself can function as a model of the world, with a structure rich enough for machines to learn from. Systems trained entirely on language are learning to navigate meaning, make inferences, solve problems—not through symbolic rules or physical experience, but through immersion in the statistical texture of text.
Summerfield never overstates this shift, but he doesn’t downplay it either. He walks carefully through how we got here—from early thought experiments to the rise of LLMs—and makes clear just how surprising this turn has been, even for those closest to the work. He reminds us that this is not just a technical leap, but a cognitive one. A threshold in our understanding of what minds are, and what they might be. I love this stuff!
In the way he writes—you can tell he wants it to be a fun read—he gives you space to think, to sit with the implications. That the tools we’ve built can now speak, reason, and engage in ways that feel familiar but also unfamiliar. He shows how the boundaries between language, thought, and world-modeling are not where we thought they were.
This is a book about orientating to a new scientific fact—that language is even more than we thought it was. It’s more than how we communicate. Language is also an astonishing model of reality, so powerful and rich that machines can learn about the world from language alone. Summerfield shows that language is now the channel through which AI achieves surprising capabilities, and this realization reorients how we see not only artificial minds, but our own. Through clear storytelling and a unique cognitive science lens, the book makes the case that language is not simply a mirror of the world, but a medium by which the world itself can be learned and modeled in ways we never imagined.
You can listen to our podcast with him from a few years ago here. And we are planning for him to return to the pod soon!
