It reads like a textbook filled with clearly digestible discussions covering AI, past, present and future.
I loved the structure. It's a mix of simple explanations of AI technologies and applications, really interesting questions, as well as clear transitions. I really loved the ‘coming up’ feature. As the chapters are short and simply written for a non-technical audience, it is engaging and readable in one sitting. I enjoyed a few cups of coffee and read this cozy in my local coffee shop.
Co-created with AI, I believe, it is in itself an excellent example of collaboration. A book that cuts the nonsense and keeps all the really good stuff.
Four chapters stood out for me in terms of learning:
- RAG vs. Fine-Tuning: The contrast between Retrieval-Augmented Generation (RAG) and fine-tuning helped me understand the architecture of AI systems and how they collect real-time information, making them more adaptable and dynamic. This chapter along with a few on explaining the technical to a non-technical audience were excellent for me. For example, I never really understood what transformers were, let alone how they worked.
- Democratising AI: The idea here was that AI systems are being made widely available, i.e., the open-source models. Similar to the point made in the research chapter, the idea is that now ordinary citizens without huge computational power can get involved in innovation, creativity and development. However, it’s worth mentioning that these technologies are available as a gift of private institutions, and that gift could be revoked at any time. It also reminded me of the suspicion I've long had and captured in the phrase, 'if it's free, your the product!' The promise of democratisation raises the question of who really controls the future of AI if access remains in the hands of private institutions.
- AI Agents: Everyone talks about agents, and I was familiar with the concept, but what I found fascinating was the idea of systems where many mini AI agents work together to create their own hive mind. AMAZING! I thought. I now understand how this increases accuracy, efficiency, and problem-solving potential by mimicking swarm intelligence. This is perhaps the future, not just agents, but swarms?
- Introducing Silicon to Cells: This chapter was a bit scary, and I wasn’t aware of SBI as a live concept. The idea of embedding technology using biology as a substrate, rather than silicon—using cells, circuits with neurons, and computers with organoids—was very odd and activated my fear response. I had no idea of the extent of this co-development. It feels like science fiction, but it’s presented as an inevitable step in the evolution of AI. I think he Godwin is correct in this assessment, and I see it as massively increasing exiting inequalities.
Overall, The New Mind balances simple explanations to allow a non-tech person to understand deep tech issues.
My only suggestion would be that the book could have taken a deeper look at the ethical concerns. While several chapters touch focus on ethics, the balance of overall risks and opportunities in this book felt skewed to the opportunity, of expanding what we think of as creativity and ultimately what we think of as human. It see perhaps more risks in this brave in world.
Thank you,
Godwin Josh
, for the pre-sale copy.
