The AI-Augmentation Revolution

The future is a centaur world

I’ve been finding more use-cases for AI in my professional life over the last couple of months. I now have ChatGPT open in one of my browser tabs more regularly than not. I ask it questions, ask it to help me rephrase things, and ask it to expound on concepts so that I can quickly get smart on a particular topic, or to help me organize my thoughts. It’s augmenting my work regularly. So today I wanted to write a bit about what I consider our future. We had the agricultural revolution that brought us cities back about 10,000 years ago, then the industrial revolution about 250 years ago that brought us wide-spread wealth and material well-being, and now we are at the dawn of the AI-augmentation revolution. To talk about this, I am going to reference two books, Tyler Cowen’s 2013 book, Average is Over, and David Goodhart’s 2020 book, Head, Hand, Heart.

Average is Over is where I first read a serious consideration of how AI would be used in the future. He reasoned from his observation of competitive chess that fairly rapidly there came a point where even the greatest chess grandmaster could not beat a computer. But then he noted that a human paired with a computer could beat a computer. These were “centaur” teams - half man, half computer. There was something a human could bring to the chess table that a computer alone could not. Things have evolved since then, and I don’t know what the state of play is now - I think it is now multiple computers fighting it out, with no humans, but who knows. It will probably be different next week. Here’s the thing: chess is a “tame” problem - it has clear rules and the rules do not change. You solve the problem within the set of rules. It is complicated, but not complex. The real world we live in is not a chess game, and the problems we face are often not bounded by clear rules that do not change. It is complex. In the real world, forces interact and the rules change unpredictably. Most of the big problems facing society cannot be solved by a single, smart person or even an AI. These problems are called “wicked problems”. 

Here’s how ChatGPT explains it:

Chess is a classic example of a well-defined problem. It has clear rules, a finite set of possible moves, and a definite objective: checkmate the opponent's king. The parameters of the game are known, and the state of the game is fully observable. While chess can be complicated and strategically demanding, it ultimately operates within a closed system with fixed rules and predictable outcomes.

Wicked problems, on the other hand, are complex, multifaceted issues that resist straightforward solutions. These problems typically involve numerous stakeholders with conflicting interests, unclear boundaries, and dynamic, interconnected variables. Examples of wicked problems include climate change, poverty, healthcare access, and social inequality. Unlike chess, there is no single "right" answer to wicked problems, and solutions are often subjective, context-dependent, and may have unintended consequences.

Chess mastery used to be a sign of a high level of cognitive processing power. It required extensive memorization of algorithms (strategies) and being able to see which one would be the best response to an opponent’s play. But ultimately with enough computing power, computers are better at this kind of cognitive work than humans. AI (machine learning) is better at pattern recognition than humans, and anything that can be boiled down to a set of rules can be programmed for a computer. Machine learning can even help find the rules in a seemingly messy process. 

Back in 2013, Cowen made the argument that AI will automate many white collar jobs that require basic pattern recognition and the following of algorithms. Much of finance and accounting, bread and butter transactional work in law (writing contracts, wills, etc.), and other fields will finally fall to automation, the way factory work did in the late 20th century. Medicine is beginning to be transformed by AI, as diagnosis and treatment are largely algorithmic. Cowen predicted that the people who will be successful in the future will be the ones who can partner with AI to do their work. AI augmentation will make traditional white collar workers far more efficient. But it will also eliminate many of the jobs in many traditional white collar fields, much the way farming and then manufacturing were automated. A farmer with a combine is far more efficient than a farmer with a horse-pulled plow. You need a tiny fraction of the human labor to generate a multiple of the product that you did before machine complementation. The cool thing is a farmer with a combine makes a much better living than a farmer with a horse-drawn plow. This is true for anyone who survives automation and becomes the operator of a new automated process. A heavy equipment operator who can operate a backhoe makes more than a man with a shovel. A radiologist who has an AI complement reading films is going to make more than a radiologist who tries to read films alone. 

There is a lot of hand-wringing that automation will leave us with a world where there is no work left for humans. Some people are calling for a uniform basic income (UBI) because people won’t have jobs. It wasn’t that long ago that the vast majority of humans were employed in agriculture. In America today that number is less than 5%, but there is no shortage of work. The transition from agriculture to manufacturing to services has not been without challenges to individuals. And the transition to whatever comes next with the AI revolution will hurt a lot of us - especially white collar workers (and probably the people like me who educate white collar workers). But the people who survive - the centaurs - will each create much more value for society. 

I’ve written before about David Goodhart’s 2020 book, Head, Hand, Heart. Goodhart argues that the modern labor market can be split into three broad categories:

Head - cognitive work focused on pattern recognition, analogic reasoning, data analysis, and decision making. 

Hand - manual work such as construction, cooking, plumbing, etc. 

Heart - emotional labor, which involves managing emotions, building relationships, and providing care and support, such as nursing, child care, social work. 

There are many fields that use some of each, such as nursing, which has components of all three, but I think it can be argued that what differentiates nursing from medicine is the focus on the heart at the center of the discipline.

In many ways the post-industrial economy has heavily rewarded “head” work over the other two, and government policy has directed investment toward head work as well. The emphasis in secondary schooling on getting to college rather than learning a trade over the last several decades is part of that shift. The fact that there is a shortage of skilled workers in the trades now and that many people with degrees are working in fields that do not require college education (and the college loan “crisis”) is a clear indication that we overinvested in head preparation. Furthermore, the coming AI-augmentation revolution is going to radically decrease the number of people doing basic head jobs. We will still have lawyers, doctors, and accountants just like we still have farmers and factory workers, but what those future lawyers, doctors, and accountants do will look quite different from today, and there will be demand for a fraction of the number. Employment in agriculture and manufacturing cratered, some 50-75 years apart, and the same thing is going to happen to white collar work.

As the demand for head work decreases, the demand for hand and heart work will increase. 

Hand - humans remain far more adaptable than robots. Robots, even with AI assist, will still be best for repetitive tasks in a tame environment. The factory that pushes through thousands or millions of units in a controlled environment is like the game of chess. Action in the real world, building, farming, etc. may not rise to the scale of wicked problems, but they will require fine adjustments that do not necessarily follow an algorithm. Robots will also need to be maintained and repaired. Many hand tasks still require adaptive solutions and do not follow algorithms.  

Heart - AI-assist is going to be useful to humans in managing transactional interactions such as customer service calls, but making genuine connection is still going to be a human strength. AI-assist will allow caregivers to focus more on high-value activity by automating recurring needs, but there will be an increasing demand for better connection. I can imagine a world in the future where humans feel truly seen and cared about as they go through their daily business because heart capacity has been freed up and enhanced by AI. 

Hand and heart work in the real world with real people. In many ways, a manager’s role is more heart (emotional intelligence) than head (cognitive intelligence). Even in white collar fields, AI-augmented workers will still need to be managed, and to work in teams. It is common knowledge in the field of management that the person best at a job is not necessarily the best person to lead people doing the job. 

I think the future world of work is a world where most humans are centaurs, working with some form of AI-augmentation. The AI-augmentation will cheapen the value of human head work, while raising the relative value of hand and heart work. Very little of the world is as controlled as a game of chess, so the AIs have a long way to go yet. Wicked problems, or even problems that do not rise to the level of wicked (maybe just bad problems? Problems of ill will?) will still require a human component that can act with intuition and adapt. 

Limitless and cheap cognitive capacity is going to cheapen the importance of cognitive skill relative to the ability to physically manipulate the environment or to form connections and motivate teams. Limitless AI-augmentation may even help us solve truly wicked problems such as climate change, poverty, and public health. I think the AI-augmentation revolution is going to bring us a world of far more wealth, as the industrial revolution did, but also more joy as we are freed to focus more on the heart. To flourish in this next revolution, we are all going to have to become centaurs.