New book: AI and Productivity by Donald Clark

The real problem of humanity, wrote the biologist E. O. Wilson, is that we have Palaeolithic minds, Medieval institutions and Godlike technology.

That line captures the heart of my new book, AI and Productivity. It is about that three-way tension between who we are, how we organise ourselves, and the tools we have built. Humans are still fragile, distractible primates trying to get through the work day. Our institutions are almost always out of date. And AI technology has suddenly become almost divine in scope. The question, then, is whether our minds and systems are remotely ready for what we’ve created.

Palaeolithic minds

Let’s start with us. Humans are brilliant in many ways but we’re also slow learners. It takes two decades to train a professional brain, even then it forgets, makes mistakes and burns out. We need to sleep, rest, retire, decline and die.! Digital minds don’t. They read millions of books in weeks, remember everything and never switch off.

We are also riddled with cognitive quirks. We cling to our beliefs (confirmation bias), stick to what’s familiar (status quo bias), and exaggerate dangers (negativity bias). We procrastinate, overthink and confuse activity with achievement. We let work expand to fill time and wait too long to act. And yet here we are, having built machines that can think. Silicon now reasons, writes and learns at a scale our ancestors could never imagine.

Medieval institutions

If our minds are limited, our institutions are hugely unproductive. We still run governments, schools, Universities and corporations on systems designed for the 19th century, sometimes earlier. They struggle to keep pace with exponential technology.

Economists have long noticed this mismatch. Robert Solow pointed out in the 1980s that “you can see computers everywhere but in the productivity statistics.” Big technologies take time to show their value. They need investment, infrastructure, skills and governance to catch up.

The paradoxes keep piling up. Jevons found that greater efficiency can increase demand, not reduce it. Friedman reminded us that if the goal were merely ‘jobs’, we could dig with spoons instead of shovels. The Easterlin paradox showed that higher incomes do not necessarily make us happier. And Peter Turchin warns that overproducing university graduates leads to frustration and instability. Legacy software, such as the LMS and primitive authoring tools still dominate online learning. Our systems aren’t designed to handle these feedback loops. They regulate too late, distribute too little, and adapt too slowly.

Godlike technology

Meanwhile, the tech itself races ahead. The past twenty years were the ‘attention economy’—social media, filters, delivery apps and endless distraction. But AI is moving the frontier from entertainment to the real economy.

Intelligence is now a utility. AI tutors can teach at scale. AI in healthcare can diagnose faster and cheaper. Agents already co-ordinate workflows, summarise research and design campaigns. And looming beyond this is AGI, machines that don’t just assist us, but act on our behalf.

Work is changing fast. White-collar jobs, once thought safe, are being automated task by task. Meanwhile, some blue-collar work, plumbing, driving, repairing, remains stubbornly human, at least for now. But even here, ‘steel collar’ robots are moving in.

Don’t forget that this is a general-purpose technology, like electricity or steam but faster, global and exponential. A billion users have arrived almost overnight. Ignoring the productivity consequences is to ignore the empirical truth.

The paradoxes of productivity

Every technological revolution has its panic. The Luddites feared looms; Keynes warned of ‘technological unemployment’. Yet jobs rarely vanish outright, they morph. The tractor ended farm labour but built the modern food economy. Computers wiped out typing pools but created the software industry.

AI feels different because it touches thought as well as labour. It automates not just muscles but minds. Generative models now draft legal briefs, write code, analyse scans, and summarise research, all in seconds. Entire workflows, not just tasks, are being handed over.

Still, new roles are emerging: AI Architects, Prompt Engineers, Data Curators, Chief AI Officers. The labour market is reshaping itself around the orchestration, governance and design of intelligent systems. The challenge is timing, whether we can reskill fast enough, and whether the new jobs carry the same pay, purpose and dignity.

And the great irony? Productivity gains do not always mean more leisure. Jevons showed that efficiency often increases total consumption. When AI helps you write reports faster, you’re simply asked to write more. The faster we work, the more work we create.

Milton Friedman put it best: if the goal is merely ‘jobs’, we might as well dig with spoons. The real measure is not how hard we work, but how much value we create and how well we use the time we save.

The real question

So the story of AI and productivity is not about utopia or apocalypse, it is about adaptation. Can our Palaeolithic minds and Medieval institutions catch up with our Godlike technology?

That’s the question this book explores. Many surprises were uncovered while writing the book, as well as clear recommendations. What is productivity (not as simple as you think)? Are our brains wired for productivity (not really)? What sectors are moving faster than others (jagged frontier)? How productive will embodied AI be (massive)? How do we implement AI in organisations (not with AI course)? What are the critical ethical issues around productivity? Where are we heading on productivity? I will be surfacing these ‘surprises’ and ‘recommendations’ in my talk on Friday 5th December, which is also my birthday!

Written for OEB 2025 by Donald Clark