🐦 Nexus

Author: Yuval Noah Harari
Fiction?: Non-Fiction
Genres: History, Politics, Technology
Rating: ⭐⭐⭐⭐
Date Finished: December 22, 2024
Notes Status: Notes In Progress

🚀 The Book in 3 Sentences

1. Humanity creates information networks in pursuit of power, order and truth in various levels of focus.
2. Revolutions in information networks have historically provided opportunities for tyranny as well as prosperity, and required significant work and error-correction to provide progress.
3. AI information networks are the first in human history to be able to make independent decisions as part of the network without intervention, opening up access to our world for this new “alien intelligence”.

🎨 Impressions

Once again an excellent book from Yuval Noah Harari blending history, politics and technology on such an important topic as artificial intelligence. I considered this book almost as a thesis in how AI will shape politics based on historical information revolutions rather than a simple discussion of AI in the world. Yuval posits a couple main hypotheses: the new “alien intelligence” AI is a notable new information network that is nothing like humanity has yet seen, and that information networks inherently connect people rather than seek “truth”. There are a few repetitive moments throughout the book where I felt a bit of a slippery slope argument and sometimes a false dichotomy, notably around the statement that AI is “evolving”. I think that AI is still at the level of human design, and while AI is already feeding in coding decisions, the overall training/data source etc. is still under human control. Hopefully with more key individuals reading this book, it will stay that way.

How I Discovered It

It’s a book by Yuval Noah Harari, it’s in the window of every bookshop.

Who Should Read It?

I think it’s an interesting albeit long read for everyone. Reading or listening to an abridged version is probably a good way to get into this book, and then follow it up with a deeper dive into the details. For a full understanding of the alarm behind Yuval’s words, it does require most of the length of this book.

☘️ How the Book Changed Me

💡 It was enlightning to read the opening arguments regarding information networks, their goals and the various goals they have.

💣 The inevitable arrow of time does not guarantee that technological progress veers towards freedom and truth, as evidenced by the moments of crises brought on by revolutions in information networks like the written word, printing press and telegram. This is important to keep in mind with the rosy-tinted messaging coming out of most tech news.

📜 I enjoyed the deeper dive into historical background as a basis for the argument as to why AI is so much different to anything that came before.

🏛️ The importance of self-correcting mechanisms and institutions for a functioning system is underscored multiple times, and reinforced this concept for me.

✍️ My Top 3 Quotes

“Ignorance is strength” (I know it’s actually from 1984 but it was important to mention it in Nexus)
“Elections establish what the majority of people desire, rather than what the truth is. And people often desire the truth to be other than what it is.”

“Just as the law of the jungle is a myth, so also is the idea that the arc of history bends towards justice.”

📒 Summary + Notes

Nexus explores how information networks have shaped human civilization from ancient times to our AI era. The book examines how these networks - from early storytelling to modern digital systems - have been instrumental in establishing power, order, and truth throughout history. Harari investigates how different societies have wielded information for various purposes and warns about the unprecedented challenges posed by artificial intelligence in our modern information networks.

The fundamental challenge remains data quality - its disorganization, inconsistency, and decentralized nature. While data collection has become easier, the critical work lies in data cleaning and standardization. Modern approaches like latent space contextual language models show promise but ultimately rely on curated data. The book effectively addresses this, particularly in explaining how AI can be directed toward specific objectives.

The “alien intelligence” paradigm provides a useful framework for understanding current AI systems. While the internal workings may be opaque, we comprehend the overall system functionality. AI’s reliance on mathematical loss functions and metrics for training ensures some level of interpretability, at least until AI potentially develops loss functions beyond human comprehension.

Historical perspective is valuable here - in the 1900s, predictions of flying cars by now proved wildly optimistic when we still struggle with basic infrastructure like sanitation globally. However, the scale and nature of potential AI developments warrant serious discussion, even if the book occasionally tends toward alarmism. Importantly, the components for both positive and negative scenarios already exist, regardless of whether we achieve superhuman AI.

Harari concludes with a pragmatic call to action:

The good news is that if we eschew complacency and despair, we are capable of creating balanced information networks that will keep their own power in check. Doing so is not a matter of inventing another miracle technology or landing upon some brilliant idea that has somehow escaped all previous generations. Rather, to create wiser networds, we must abandon both the naive and the populist views of information, put aside out fantasies of infallibility, and commit ourselves to the hard and rather mundane work of building institutions with strong self-correcting mechanisms. That is perhaps the most important takeaway this book has to offer.

I highly recommend a thorough read to get the really interesting historical background to this conclusion.