Book review: A Brief History of Intelligence

Author Max Bennett explains to JP O'Malley why the evolution of the brain holds the key to the future of AI

Max Bennett takes readers back over four billion years to track the evolution of the human brain and the creation of AI.

• A Brief History of Intelligence: Evolution, AI, and the Five Breakthroughs That Made our Brains 
• Max Bennett 
• William Collins, €17.99 

In 1950 British mathematician Alan Turing published 'Computing machinery and intelligence'. The scientific paper contained a thought experiment known as the “Turing Test”: This proposed that a computer can be said to possess artificial intelligence (AI) if it can mimic human responses under specific conditions.

Machine learning has come on in leaps and bounds in the interim. Today, AI can beat its human counterpart in a game of chess; read radiology images better than its human counterpart; and is currently on the cusp of autonomously driving cars.

ChatGPT, meanwhile, can compose poetry, translate between languages at will, and even write code.

Still, machine learning has many limitations. Max Bennett is still trying to figure them out.

The youthful and successful American entrepreneur cut his teeth in the business world as a trader at multinational investment bank, Goldman Sachs. Currently, he is the co-founder and CEO of the AI company, Alby.

10 years ago, Bennett co-founded Bluecore. The company is valued at $1bn (€919.6m) and uses AI to assist leading global brands with their marketing strategies.

Discovery of a gap in the market

Bennett came up with the idea for the company when he discovered there was a gap in the market.

“Amazon was doing a lot of cool stuff with machine learning to help predict what customers wanted and to help personalise marketing based on their behaviour, which the rest of retailers and brands [around the globe] didn’t have access to,” Bennett explains.

Bennett has since moved to an advisor role in Bluecore. For the past two years, he has been travelling the world and working on the manuscript of A Brief History of Intelligence: Evolution, AI, and the Five Breakthroughs That Made Our Brains.

“Why is it that AI can crush any human in a game of chess, but can’t load a dishwasher better than a six-year-old?” the author asks.

Even the most sophisticated AI programmers struggle to answer this question, the American tech entrepreneur explains.

“What AI programmers find easiest is recreating a system’s ability to do math or to use language,” he says.

Recreating intelligence that evolved way earlier though — like, for instance, moving around in continual action space is much more difficult.

Bennett’s thoroughly researched, ambitious book provides a detailed account of how these primitive features of intelligence evolved.

It’s a huge undertaking of both science and history, chronicling how the human brain evolved over the last 4bn years.

Bennett doesn’t offer any new original insights as such. He mostly cites the work of older and wiser experts — including evolutionary biologist Richard Dawkins, philosopher Daniel Dennett, neuroscientist Antonio Damasio, and psychologist Daniel Kahneman.

Still, he organises that complex and intriguing evolutionary history into a lucid, reader-friendly narrative for the non-specialist.

The human brain has 86bn neurons and over 100tn connections. To state the obvious: It’s complicated.

To really understand the brain, we need to understand its history, Bennett says.

We, therefore, need to examine the fossil record of the history of life. The book is divided into five “major breakthroughs” stages.

The first 3bn years of life on Earth existed without brains. “For most of the time life existed on Earth, it was all single cell organisms,” Bennett says.

Then, roughly 600m years ago, radially symmetric neuron-enabled coral-like animals reformed into animals with a bilateral body.

British mathematician, logician, cryptanalyst, and computer scientist Alan Turing is widely considered to be the father of computer science and artificial intelligence.

“In this very first brain came the early affective template of animals: Pleasure, pain, satiation, and stress,” Bennett says.

“Most researchers also believe that associative learning first emerged in bilaterians.”

500m years ago, one lineage of ancient bilaterians grew a backbone, eyes, gills, and a heart — becoming the first vertebrates (animals most similar to modern fish).

Their brains formed into the basic template of all vertebrates: A cortex to learn, to recognise patterns, and to build spatial maps and the basal ganglia to learn by trial and error.

Sometime around 100m years ago, in a four-inch long small ancestral mammal — the dorsal cortex of our ancestral vertebrate — reformed into the modern neocortex. 

This neocortex enabled animals to internally render a simulation of reality. It gave them the power to learn by imagining and the ability to plan for future events.

“Neuroscientists are fascinated with the neocortex — a sheet that is bundled up surrounding the centre of the brain — because it seems to be the epicentre of human intelligence and encompasses counterfactual learning, episodic memory, and imagination,” Bennett adds.

If you look at a human brain, most of what you see is neocortex.

Sometime around 10m-30m years ago, new regions of neocortex evolved in early primates that built a model of the older mammalian areas of neocortex.

This, in effect, meant that these primates could simulate not only actions and stimuli (like early mammals), but also their own mental states with differing intent and knowledge.

These primates could then apply this model to anticipating their own future needs, understanding the intents and knowledge of others (known as theory of mind), and learning skills through observation.

“There is pretty weak evidence suggesting that theory of mind exists outside of primates,” Bennett says.

“Maybe something exists like this in really intelligent mammals — like dolphins or elephants — though.”

Evolution of language

The brain’s fifth breakthrough came via the evolution of language. How it evolved still baffles scientists and historians alike.

The philosophical conundrum poses more questions than answers. Namely: Did the brains of early humans (over 2m years ago) gradually evolve some unique ability to speak?

Bear in mind, language vocal chords only seemed to have evolved around 500,000 years ago.

Or is language just a cultural invention that occurred over 50,000 years ago and was passed down through generations?

“Because we have no intermediate steps as to how and why language evolved, we are kind of guessing,” Bennett says.

Paraphrasing the work of the Israeli historian, Yuval Noah Harari, Bennett notes how language — combined with technology — has given our species divine-like powers to build civilisations, invent religions, create art, and construct AI in our own image.

Bennett claims evolution is still unfolding in earnest. He points out that the universe is only 14bn years old.

Life on Earth, meanwhile, is only 4bn years old. What will the next step be for our species look like?

“My high confidence prediction is that we are going to have AI systems with human level intelligence, instantiated in silicon, within our lifetime,” Bennett says.

“There might be people who challenge whether these AI systems are conscious or not, but it will be almost impossible to tell the difference between a [human] and [a machine] when interacting with one of these things.”

Will AI systems have human emotions though? Like shame, for example?

“It’s really up to the designers of these systems, if we want to give them human-like emotions,” Bennett says.

But there is a great danger in giving AI systems these human traits.

“After all, it’s these traits that have led [our own species] to become so tribal, giving us all the bad stuff in humanity — like devastating wars.

“I think if we have learned anything from ChatGPT it’s that there is not going to be this moment where we wake up, like in the movies, and say: ‘Oh my God, artificial intelligence has been created’,” Bennett concludes.

“Will we actually have robotic AI systems, walking around that are human-like? It’s an open-ended question. Regardless, though, in the next ten years our lives are going to be fundamentally transformed by AI.”