Abstract
Most famously Frey and Osborne’s (2017, p. 254) inquiry “How susceptible are jobs to computerization?” has fundamentally questioned whether the market-based economy in its current form will remain a satisfying solution as a wealth distribution mechanism. Neo-Luddites instigate fears that recent and predicted progress in data processing, machine learning, artificial intelligence, and robotics may challenge the viability of most blue- and white-collar jobs.
Thomas W. Malone in his book Superminds does not entertain such fears, but instead focuses on the idea of how people and computers achieve superiority by combining their abilities. At the core of his argument, he shows how computers enable humans to overcome coordination challenges. He attests much potential to make smarter decisions and to tackle some of the grand challenges that humankind is facing.
The book is divided into seven parts with two to six chapters each. Instead of following this structure, I discuss three primary insights that Malone develops and carries through the book as a way to fairly represent the work in a comprehensive manner and to identify the core ideas of the book.
Superminds – a group of individuals acting together in ways that seem intelligent (p. 20) – have shaped human history. Malone categorizes four groups of superminds, i.e., hierarchies, communities, markets, and democracies, to which he refers many times in the book to exemplify how decisions are enabled or coordinated through computers.
His arguments rest on the assumption that groups outperform individuals in terms of intelligence and problem-solving. The book introduces the world of hyperconnectivity and intelligence through the example of the Polymath project in 2009 (p. 5), which was an open online-based project to prove the Hales–Jewett theorem. With the help of 39 people and 1500 comments, in two months the proof was completed.
Computers constitute minds that may add to decision-making. Computers increasingly outperform humans in specific tasks that have clear boundary conditions. In turn, humans are much better general problem-solvers as they can spot the nature of issues more unerringly and thus apply appropriate algorithms to a wide range of issues. For instance, a chess computer may beat a human chess player, but the same computer is likely worse in playing Mikado. This idea connects well with the notion of hyperspecialization developed in the book. Hyperspecialization does not categorize the division of physical labor, but the one in data collection and analysis to gain more and better information. The division of labor through hyperspecialization may announce another age of huge productivity gains. Hereby, computers are not substitutes, but complements to human activity.
Malone develops a second set of ideas around the question of how people and computers already cooperate in daily life and will increasingly do so in the future. According to him, the primary function of artificial intelligence in the next two decades will be communication. Communication has evolved fundamentally in the last thirty years through computers and, enabled by ever smarter machines, takes on a variety of new forms. Automated assistants govern more and more of our daily work. For example, if we need information about something, we are likely to use a highly sophisticated algorithm called Google. Machines become peers that fulfill the same function as humans, such as in self-driving cars or automated trading systems. Machines become managers in the form of traffic lights and many user-based systems that coordinate the work of crowds like the Crowd-Forge system (p. 55). We have benefited from these arrangements and will likely benefit even more in the future.
Third, Malone devotes significant space to the idea of governance of decisions that he later leverages to exemplify some of the areas in which he expects large changes to happen. The discussion evolves around the five governance systems: hierarchies, democracies, markets, communities, and ecosystems. The governance systems leverage the gathering and interpretation of information (called sensing in the book), improve memory, for example by structuring symptoms of medical databases for better diagnosis, and increase learning through pattern recognition. While humans will always be final decision-makers in each of these scenarios, their decisions will systematically be informed through knowledge bases with analytical capabilities.
In the remainder of the book, Malone provides examples about areas in which superminds could be particularly effective, what risks artificial intelligence poses, and whether we should consider superminds to be conscious (the answer in many cases is yes).
The book is based on the unique path that Malone has taken as a professor of both management and information technology, having also founded four start-ups in the IT sector. Superminds is an original idea and one of the first comprehensive conceptualizations of mindfulness and consciousness on an aggregated level. One of the big topics in management research is the question of how we can aggregate our understanding from the individual level to higher levels and accurately describe and possibly even predict collective motivations, thoughts, behaviors, aspirations, and outcomes. Superminds is a book that offers such a model. The strength of this model is its humility that points out boundary conditions and limitations as to what extent we can understand individuals as superminds. At the same time, this is also a weakness as much more work will be required to answer the ‘so what’ question to some of the ideas that Malone outlines. Much of the potential implications remain relatively vague speculations and hopes rather than directions to move forward and scenarios of how the increased adoption of artificial intelligence can and possibly should be developed.
Part 6 on risks of artificial intelligence remains superficial, addressing popular concerns in the way that doesn’t reflect the thoughtfulness of many other chapters of the book. One of the reasons for this perception may come from the time horizon that the author applies. Most of the time, the reader is left alone whether we imagine a world in ten or fifty years. Admittedly, Malone states early on that he expects most relevant changes in the next twenty years to take place in the degree of integration and support of communication tools rather than in any other area.
I am and have always been fascinated by the possibilities of new technology. This is why I enjoyed reading this book. The concept of superminds is fascinating. Malone has a very positive outlook and focus on the opportunities that this concept may open up. This endeavor deserves commendation.
However, I sometimes felt that the summary of research fields was favorable to his argument and not always as balanced and inclusive as it could have been. While there is much intent to display an international mindset in the book, it remains clear throughout that the perspective taken in this book is USA-centered. Real superminds, however, may not be. As a consequence, I miss some depth in the second half of the book that shapes the tenets of the fascinating ideas of superminds further.
One approach that might have mitigated this impression would be a more balanced narrative of benefits and potential costs. Possibly, it is simply too early to paint such a picture. Cybersecurity and data security have been one of the central tenets of the European high-tech scene. Solutions for these concerns are not yet easy to scale, but possibly, in a post-pandemic world, responsible management practices may emerge as something more central. In some way, these concerns have already become louder in the context of international interference in elections and cybersecurity risks in relation to national security. The book remains uncomfortably silent about opportunities in this area.
The most intriguing claim of the book is that our enhanced data processing and analytical capabilities allow us to deploy new forms of governance that empower individuals and weaken traditional hierarchies. This idea may challenge the utility of some of our decision-making and problem-solving theories in their very assumptions (e.g., Nickerson & Zenger, 2004).
Fleming (2019) pleads for an answer to the question of how we can design our interaction with AI in a way that indeed empowers individuals rather than marginalizes them. Admittedly, this question remains unanswered in the book over the excitement of new opportunities from superminds consisting of the combination of humans and individuals. Possibly, a reference to sociomateriality (Orlikowski & Scott, 2008), with more attention to the fusion of technology and work in organizations, could have strengthened this large missing piece. I may add, learning from recent interviews by some of my dissertation students with managers, that many managers are not aware of the extent to which AI is already integrated and embedded in organizations.
Nowadays, neither are individuals able to see the full complexity of issues, nor is it necessary to have single decision-makers. Interconnectivity and hyperspecialization offer solutions that likely exceed those we have been able to make in times of the concentration of information in a few hands. Malone formulates his dream that democracies can truly be lived through the real-time support of AI, and concentrated decision power may become more and more outdated as there is less and less need to concentrate information and thus decisions for efficiency gains.
In sum, Malone develops a brilliant concept that provides us with much food for thought. He is an inspiring writer and takes the reader through the world of social sciences drawing from an insightful set of theories of psychology, management, organizational behavior, political sciences, and information management. Superminds is a thought-provoking journey that envisions many exciting opportunities. A book not to be missed out on the reading list.