Will AI redefine what it means to be human?

Nothing exemplifies our anti-human age better than the fears over artificial intelligence. AI is expected to take our jobs, dominate the world and redefine what it means to be human. Blood-and-flesh people are ultimately predicted to take a backseat in our own history. This atmosphere of widespread misanthropy makes the publication of Neil D Lawrence’s book, The Atomic Human: Understanding Ourselves in the Age of AI, both timely and significant.

Lawrence, currently the DeepMind professor of Machine Learning at Cambridge University and former director of machine learning at Amazon, is uniquely placed to examine AI’s practical and existential implications. The Atomic Human explores the fundamental question of what it means to be human in an era when AI is increasingly capable of performing tasks once thought to be exclusive to human intelligence.

The book’s central thesis revolves around the concept of the ‘atomic human’, a metaphor inspired by Ancient Greek philosopher Democritus’s theory of atomism. Democritus posited that matter could not be divided infinitely but eventually reached an indivisible ‘atom’. Lawrence similarly argues that AI’s ability to replicate various facets of human intelligence will reveal an irreducible core of our humanity. This core, Lawrence suggests, is characterised by our vulnerabilities and limitations.

The Atomic Human is interesting and engaging, using rich analogies and examples drawn from history. Lawrence is, to his credit, a defender of the uniqueness and beauty of human intelligence. He correctly suggests that machine intelligence could be a beneficial adjunct to human intelligence, provided we understand and seek to control the power of these machines. These tools, as he puts it, serve as a supplement to human decision-making and problem-solving. In short, Lawrence attempts to defend human agency – human beings are the subject, not the object of change, he argues.

However, the book’s argument that human intelligence is defined primarily by its limitations, rather than capabilities, is problematic. It reflects a diminished view of personhood that is widespread in the 21st century.

Lawrence uses the example of Jean-Dominique Bauby’s The Diving Bell and the Butterfly to illustrate these human limitations. Bauby, the former editor of Elle magazine, suffered a massive stroke that left him with complete paralysis except for his left eyelid. Despite being mentally lucid, he was trapped inside his own body. Yet he was able to write his memoir by blinking to select individual letters. The diving suit represented how it felt to be restricted in this way. The butterfly represented the freedom of his internal thoughts and memories.

This ‘locked-in’ syndrome is how Lawrence characterises the human condition. Our intelligence, he says, is also heavily constrained in its ability to communicate. In this sense, we are all butterflies within a diving suit, especially when we contrast this with the speed at which the machines we have created can exchange information – tens of millions of times faster than humans.

For Lawrence, this is critical to understanding the nature of our intelligence. Machines can already replace us physically – think of combine harvesters, mechanical looms and robots building cars. AI can now displace us in mental work, too. As Lawrence says, AI’s ‘capacity to consume and process… information renders the printing press, so vital to our development, a laughable anachronism’.

Curiously, though, Lawrence does not address the obvious counterpoint here – that in creating AI, humanity is beginning to solve the problem of our limited bandwidth. Indeed, how we compensate for our ‘locked-in’ nature is precisely what makes us human. We are constantly overcoming our biological constraints. Humans are able to fly not because we grew wings, but because we created machines to do what we could not. Contrary to Lawrence’s key argument, modern man is not defined by his limits, but by his ability to break free of them. In contrast, no computer yet conceived can or ever could escape its physical constraints in the same way.

In any case, comparisons between human intelligence and AI are always misleading. Human brains do not simply process information as computers do. The mind is not just software running on the hardware of the brain. Human intelligence is an ever-expanding, ongoing, open-ended and fundamentally social process. It is rooted in the cultural practices, emotions, morals and knowledge we have created and passed on over generations.

Nowadays, the risible idea that the unconscious machines we have created will soon come to dominate us is tragically widespread. This is more a symptom of our anti-human zeitgeist than it is proof for AI’s ‘superior’ intelligence. It is a product of a deterministic, fatalistic culture of lowered expectations, in which human beings are no longer viewed as autonomous, history-making subjects

Lawrence ends his book by reminding us of the lesson of Goethe’s The Sorcerer’s Apprentice, in which the hapless apprentice deploys a spell but cannot control the results. It is a good analogy of how Big Tech companies can sometimes deploy software systems they cannot control. But it does not have to be this way.

We need to resist the attempts to portray human beings as fragile, limited and ‘atomic’ – and as no match for the technology we ourselves have created. We can, should we choose to, shape technology to meet our needs and wishes. We must once again make the case for human beings.

The Atomic Human: Understanding Ourselves in the Age of AI, by Neil D Lawrence, is published by Allen Lane.

Dr Norman Lewis is a writer and visiting research fellow at MCC Brussels. His Substack is What a Piece of Work is Man!.