What is important about this book
The effects that AI will have on our societies are the hardest to predict. The social impact of AI depends not just on the nature of technology, but also on how it is used by people, and how this shapes the way that we treat each other. In the fifteenth century, the invention of the printing press burst onto the scene, ushering in an era of widespread dissemination of ideas. Over the past thirty years, the rise of the internet has made an immense share of human knowledge accessible, giving anyone the chance to navigate a universe of half-truths, bitter commentary, and indecipherable memes. Back then, in that more straightforward time, knowledge expanded through reading, listening, and observing, and was shared via speech, writing, or typing. At that time, human beings alone were the custodians of all that was known. Yet, quite astonishingly, that world is now fading before our very eyes.
We are moving into a future where artificial intelligence systems not only preserve humanity’s collective memory, but also analyze it, producing new ideas, theories, and creations — capabilities that, until recently, belonged solely to human beings. This technological revolution may represent not just a leap in quantity, but also in quality. AI is advancing at breakneck speed, provoking both trepidation and exhilaration.
Christopher Summerfield is a prominent figure in the field of cognitive neuroscience and artificial intelligence. He is currently a professor of Cognitive Neuroscience at the University of Oxford, where he leads the Human Information Processing (HIP) Lab, a laboratory that studies human decision-making mechanisms from behavioral, computational, and neural perspectives.
Quotes
- “Those who have knowledge, don’t predict. Those who predict, don’t have knowledge”.
- “What role will humans still play in a world where AI systems generate and share most knowledge on our behalf?”.
- “What happens if large volumes of AI-generated content – news, commentary, fiction, and images – come to dominate the infosphere?”.
- “Many users who interact with chatbots fail to understand that they are talking to an AI. The advent of multimodal AI, in which facial expressions and voice can be seamlessly combined to create a plausible video stream, is already inaugurating the next steps in the so-called digital companionship”.
- “Is it ever OK for an LLM to deliberately imitate a named individual?”.
- “In the future, people will want AI systems that can take actions in the real world”.
- “Unlike humans, current LLMs are not directly endowed with a purpose. They appear to us rather passive and limp”.
- “Over the course of history, many thinkers and writers have mused about the idea of a universal text, a giant document or library in which is transcribed everything that people could possibly say”.
- “Even for us humans, reasoning prowess in specialized domains is not born overnight. Rather, it is the result of years of hard grind, usually under the guidance of expert mentors”.
Structure and contents of the book
The book, structured into 43 chapters (excluding the introduction and afterword), is divided into six parts ranging from the origins and functioning of language models, to their ability to think, speak, and act, and finally to the most debated prospects for the future:
- How did we get there?
- What is a language model?
- Do language models think?
- What should a language model say?
- What could a language model do?
- Are we all doomed?
Summerfield makes complex concepts accessible, weaving rigorous explanations together with vivid, relatable anecdotes. From the distinctions between Descartes and Locke, to the comparison between mathematical and language-based models of the mind, and even to lighter moments such as an AI-generated family menu, each step becomes a rung leading to the heart of the book: the language models at the core of much contemporary AI research. Like the most engaging thinkers, the author takes the reader on a long journey, broken into short and enjoyable stages, leaving them with concrete tools and new questions with which to view the future of AI.
Instructions for reading this book
These Strange New Minds is a book aimed both at experts or enthusiasts of tech and AI, and at non-experts: each argument is well introduced with information that helps to understand the context, and the details provided are thorough and comprehensive. It is an essential work for anyone in business or consulting who wants to make informed decisions about the adoption, governance, or use of generative AI. In this book, Christopher Summerfield clearly retraces the historical evolution of AI, presenting a clear, insightful, and profoundly human examination of the boundaries between human and machine intelligence — revealing what these frontiers can teach us about our own minds. It’s a must-read for anyone seeking to grasp not only the direction AI is taking, but also what it reflects back about who we are.
The author notes that AI is already being used both to create value and to cause harm, with the past year seeing a sharp rise in financial fraud and intimate image abuse. To steer its development toward the common good, he calls for close coordination between researchers, developers, and regulatory authorities, alongside research programs aimed at anticipating and mitigating emerging risks. Priorities include reducing bias and toxic content, preventing dangerous uses such as the creation of biological weapons or cyberattacks, and protecting the most vulnerable — from children to those facing mental health challenges — especially when they turn to AI systems for personal or health advice. At the same time, he reminds us that LLM cognition is — and likely will remain — fundamentally different from our own. While these systems excel at reasoning through formal problems in logic, mathematics, and coding, they lack the motivation, planning abilities, and, above all, the visceral and emotional experiences that make us human: they have no body, no friends, and never feel hunger, loneliness, or frustration. They are, in a sense, minds — strange new minds — unlike anything humanity has ever known.