After the first couple of chapters, I was ready to give this book 1 or 2 stars. There was a lot of talk about artificial intelligence, machine learning and big data - all topics that are familiar to me because of my profession as a software engineer. I got the impression that the author did not fully understand what he was writing about. The narrative in those chapters is pervaded by pessimism, and paints a world where people have no jobs, and all their movements and thoughts are controlled and predicted by "the algorithm", an ominous presence that nobody really understands. In this future reality, according to the author, people would become irrelevant and would be completely dominated by machines and by a biologically-enhanced human elite. But here's a big inconsistency in this narrative: if people become so irrelevant, why is it so important for algorithms to be able to know and predict people? In other words, why should an "elite" develop algorithms that study people, if people are so irrelevant? So much for irrelevance - the "elite" still wants a lot from the rest of humanity! It sounds to me like people are just as relevant as they are today, or else "the algorithm" and the marketing machines would not give people so much attention.

Besides, nobody really knows what is going to happen with AI. How do we know that the current trend of deep learning techniques will keep evolving and will reach such levels of sophistication that will push people into irrelevance? Deep learning is just a statistical method that is enhanced by today's powerful CPUs and GPUs, but how do we know its capabilities will not plateau in the future? Today's neural networks are great at specific tasks, but they are absolutely nowhere near a position where they can match biological intelligence when it comes to things such as having a morality or conscience (mostly because we don't even know how it works with us in the first place). On the other hand, these algorithms do have the potential to enhance our lives by aiding us in mundane tasks, or in tasks where our feelings might be counter-productive (e.g. driving).

Harari only writes about pessimistic scenarios that are borderline dystopian - he should consider more optimistic ones instead. There's plenty that AI and big data can help us with.

The next few chapters were a bit better, but overall felt a bit disconnected. There was a lot of jumping here and there, and at this stage I felt the book lacked a coherent structure.

Luckily, Harari writes very well. His writing style is engaging and clear and it kept me reading despite my disagreements on the book's content and apparent lack of a cohesive narrative.

Well, I am glad I kept reading, because the end of the book is much better - especially parts 4 and 5. I guess this is Harari's familiar territory: here, he is able to combine his deep knowledge in history and sociology with clear insights about contemporary society and about human beings. The 20th chapter, "Meaning", was probably the best of the entire book.