Category:

The Five Breakthroughs

'Let's start with Artificial Ratlevel intelligence (ARI), then move on to Artificial Catlevel intelligence (ACI), and so on to Artificial Human level Intelligence (AHI).' —YANN LECUN, HEAD OF AI AT META

We have a lot of evolutionary history to cover—four billion years. Instead of chronicling each minor adjustment, we will be chronicling the major evolutionary breakthroughs. In fact, as an initial approximation—a first template of this story—the entirety of the human brain's evolution can be reasonably summarized as the culmination of only 'five' breakthroughs, starting from the very first brains and going all the way to human brains.

These five breakthroughs are the organizing map to our book, and they make up our itinerary for our adventure back in time. Each breakthrough emerged from new sets of brain modifications and equipped animals with a new portfolio of intellectual abilities. This book is divided into five parts, one for each breakthrough. In each section, I will describe why these abilities evolved, how they worked, and how they still manifest in human brains today.

Each subsequent breakthrough was built on the foundation of those that came before and provided the foundation for those that would follow. Past innovations enabled future innovations. It is through this ordered set of modifications that the evolutionary story of the brain helps us make sense of the complexity that eventually emerged.

But this story cannot be faithfully retold by considering only the biology of our ancestors' brains. These breakthroughs always emerged from periods when our ancestors faced extreme situations or got caught in powerful feedback loops. It was these pressures that led to rapid reconfigurations of brains. We cannot understand the breakthroughs in brain evolution without also understanding the trials and triumphs of our ancestors: the predators they outwitted, the environmental calamities they endured, and the desperate niches they turned to for survival.

And crucially, we will ground these breakthroughs in what is currently known in the field of AI, for many of these breakthroughs in biological intelligence have parallels to what we have learned in artificial intelligence. Some of these breakthroughs represent intellectual tricks we understand well in AI, while other tricks still lay beyond our understanding. And in this way, perhaps the evolutionary story of the brain can shed light on what breakthroughs we may have missed in the development of artificial humanlike intelligence. Perhaps it will reveal some of nature's hidden clues.

Me

I wish I could tell you that I wrote this book because I have spent my whole life pondering the evolution of the brain and trying to build intelligent robots. But I am not a neuroscientist or a roboticist or even a scientist. I wrote this book because I wanted to read this book.

I came to the perplexing discrepancy between human and artificial intelligence by trying to apply AI systems to real-world problems. I spent the bulk of my career at a company I cofounded named Bluecore; we built software and AI systems to help some of the largest brands in the world personalize their marketing. Our software helped predict what consumers would buy before they knew what they wanted. We were merely one tiny part in a sea of countless companies beginning to use the new advances in AI systems. But all these many projects, both big and small, were shaped by the same perplexing questions.

When commercializing AI systems, there is eventually a series of meetings between business teams and machine learning teams. The business teams look for applications of new AI systems that would be 'valuable', while only the machine learning teams understand what applications would be 'feasible'. These meetings often reveal our mistaken intuitions about how much we understand about intelligence. Businesspeople probe for applications of AI systems that seem straightforward to them. But frequently, these tasks seem straightforward only because they are straightforward for 'our brains.' Machine learning people then patiently explain to the business team why the idea that seems simple is, in fact, astronomically difficult. And these debates go back and forth with every new project. It was from these explorations into how far we could stretch modern AI systems and the surprising places where they fall short that I developed my original curiosity about the brain.

Of course, I am also a human and I, like you, have a human brain. So it was easy for me to become fascinated with the organ that defines so much of the human experience. The brain offers answers not only about the nature of intelligence, but also why we behave the way we do. Why do we frequently make irrational and self-defeating choices? Why does our species have such a long recurring history of both inspiring selflessness and unfathomable cruelty?

My personal project began with merely trying to read books to answer my own questions. This eventually escalated to lengthy email correspondences with neuroscientists who were generous enough to indulge the curiosities of an outsider. This research and these correspondences eventually led me to publish several research papers, which all culminated in the decision to take time off work to turn these brewing ideas into a book.

Throughout this process, the deeper I went, the more I became convinced that there was a worthwhile synthesis to be contributed, one that could provide an accessible introduction to how the brain works, why it works the way it does, and how it overlaps and differs from modern AI systems; one that could bring various ideas across neuroscience and AI together under an umbrella of a single story.