AI is creating a new form of 'co-intelligence' in which human and machine minds jointly shape thinking and production, promising major gains but demanding fresh ethical standards and governance to manage displacement, inequality and institutional risk.
The world is passing through one of the most significant technological changes ever. Artificial intelligence, which was just yesterday the preserve of science fiction, and the work of research laboratories, has become an partner in our everyday activities: it dictates our emails, diagnoses our diseases, educates our young children, controls our budgets, creates our artworks, and influences the policies made by governments and corporations that make this world the world we know it. The rate of this transition is astounding. The internet had to cope with more or less a decade before it could reach one billion users; social media did it in half times. There were now a hundred million ChatGPT users in two months. Less than a year after its debut, hundreds of millions of individuals on all seven continents were using large language models, in virtually every field of professional activity, and in most languages. None of the past technologies have spread into so many aspects of human life, so fast. However, at this breakneck pace, answers to the fundamental questions posed by AI are not novel. The question of the connection between the spheres of intelligence and personhood is one of the most ancient in philosophy and political theory, and ethics: What is the relationship between intelligence and personhood? In what cases ought we to follow the wisdom of other men or machinery? What is the way of distributing power and benefit in a society when a new and a transformative technology has come? What is our responsibility to the people that will be displaced, disadvantaged, or harmed due to change of technology? This book is not a technical manual. It will not show you how to create a neural network or instantiate a prompt, which will be the most useful to an AI system. It is a question about what it is like to be human in a world that grows more and more the co-inhabitation of intelligent machines a world where the distinction between intelligent and artificial thinking is not only becoming increasingly unclear, but also more fruitfully and more disturbingly so. Our exploration is anchored by the concept of co-intelligence, which is an imported and extended concept of Ethan Mollick in a 2024 seminal work. The concept of co-intelligence does not just refer to the utilization of AI as a tool. It talks about a new cognitive ecology where the human and artificial minds mutually influence one another to come up with ways of comprehending, creating and making choices that neither of them could accomplish individually. The intelligence that is born on the border of a human soul and a machine, truly sources of collaboration, truly radically transformative, and truly novel, is what stands. It is not whether we will change as a result of AI, it is how we will decide to transform. - Shoshana Zuboff, the Age of Surveillance Capitalism (2019) The book has six sections which deal with various aspects of the co-intelligence challenge. Part I follows the philosophical and intellectual roots of co-intelligence, delving into the history of human mental extension by use of technology and the development of AI as its ultimate manifestation so far. Part II explores the cognitive aspects of human-AI interaction what occurs, both neurologically and psychologically, during human and artificial minds unite. Part III imagines the social, economic, and political impacts of living in an AI-coexistence the impacts on work, inequality, democracy and power. Part IV aims to understand the ethical construct to implement responsible co-intelligence the issues of alignment, fairness, and accountability. Part V explores the domain-specific application in the fields of medicine, education, law and the arts, and follows the potential transformations as well as the unique dangers of AI in each area. Part VI looks ahead, providing structures of governance, and of the fostering of co-intelligence as truly fruitful to human prosperity. References are interspersed and a complete bibliography is given after every chapter and a list of master references on the last page of the volume. In cases where empirical results are in dispute, we indicate the controversy. Where there is a difference of position of these schools of thought we give them as accurately as is possible. Our aim has been to produce a book that will be accessible to general readers, and rigorous enough to be of use to specialists - a very hard feat to accomplish, and surely one which we have at one time or another not accomplished perfectly well. We have attempted to pen in neither of those two optimist/pessimist camps which are the coin positions of the popular AI discussion. We believe that the opportunities of AI in human good are real and vast; and we believe that its opportunities in human ill, in human society, in human institutions of government, and in the longer term in the environment in which humanity thrives are real and underestimated. It would be, we think, an honest position, to maintain that this is a matter of serious, engaged ambivalence: committed to manifesting the benefits of AI, but filled with concern about its dangers, and with beliefs that both the dangers and the benefits are much more a matter of human decision than of AI capacity. The machines are increasingly becoming competent. It remains up to us to make the picks.
Summary
Main Finding
The book argues that the relevant locus of change is not “AI as a tool” but a new cognitive ecology — co-intelligence — in which human and machine minds mutually shape cognition, choice, and production. This co-intelligence will be radically transformative across work, politics, and culture; its net effects (welfare, inequality, power distribution) are contingent on human decisions about governance, institutions, and design rather than on AI capability alone.
Key Points
- Rapid diffusion: Modern generative AI has spread at unprecedented speed, embedding into many daily and professional activities within months.
- Co-intelligence (conceptual anchor): Borrowing and extending Ethan Mollick (2024), the book presents AI-human interaction as mutual augmentation and hybrid cognition, not merely instrument use.
- Six-part structure:
- Philosophical/intellectual history of technological extension of mind.
- Cognitive and neurological dynamics of human–AI interaction.
- Social, economic, and political impacts (work, inequality, democracy, power).
- Ethics and responsibility (alignment, fairness, accountability).
- Domain-specific transformations (medicine, education, law, arts) and sectoral risks.
- Governance frameworks to steer co-intelligence toward public prosperity.
- Normative stance: The authors take an “engaged ambivalence” — recognizing large positive potentials and substantial risks, arguing outcomes depend on policy and institutional choices.
- Evidence and rigor: The book is not a technical manual; it relies on interdisciplinary scholarship, historical and philosophical analysis, case examples, and a curated bibliography. Contested empirical claims are flagged and debates are presented where they exist.
- Core normative questions emphasized: personhood and agency, distribution of benefits and harms, duties to displaced or harmed people, and democratic accountability in the age of intelligent machines.
Data & Methods
- Methodological approach: Primarily conceptual, interdisciplinary synthesis — combining philosophy, cognitive science, political theory, ethics, and social-science literature.
- Evidence types: Historical analysis, literature review, theoretical argumentation, illustrative case studies across domains (medicine, education, law, arts), and citations to empirical work where available.
- Transparency about uncertainty: The book identifies empirical controversies and presents competing schools of thought rather than definitive empirical claims when the evidence is disputed.
- Not an empirical monograph: It does not primarily present new econometric or experimental data; instead it frames research questions, compiles existing findings, and proposes conceptual frameworks (co-intelligence) to orient further empirical work.
- Reference apparatus: Chapter-level bibliographies, master reference list; useful starting point for tracing empirical sources and contested results.
Implications for AI Economics
- Research agenda and measurement
- Distinguish complementarity vs. substitution: estimate how different AI technologies interact with worker skills and tasks; measure changes in task allocation within jobs.
- Track diffusion speed and adoption channels: model extremely fast diffusion dynamics and network/platform effects for AI tools.
- New microdata needs: matched employer–employee data with AI tool usage, platform logs, task-level time-use, and longitudinal administrative outcomes.
- Causal methods: RCTs and field experiments of AI tools in firms, schools, and healthcare; quasi-experimental adoption studies; IV/DiD exploiting rollout or policy variation.
- Structural and macro models: incorporate co-intelligence complementarities into growth and inequality models; estimate potential long-run productivity and distributional effects.
- Welfare and distributional metrics: measure consumer surplus from AI, producer rents (platform/AI rents), and distributional shifts across income/skill groups.
- Labor markets and policy
- Retraining and human capital policy: prioritize scalable retraining aligned to AI-complementary skills; evaluate effects of upskilling on labor share and mobility.
- Social insurance and redistribution: reassess unemployment insurance, wage insurance, and forms of redistribution (e.g., wage subsidies, UBI) in light of rapid displacement risks.
- Taxation and rents: consider taxing AI-generated rents (platform profits, data rents) and using revenues for redistribution or public goods.
- Market structure and competition
- Platform concentration: anticipate stronger winner-take-all dynamics due to data and model scale; examine implications for competition policy and antitrust enforcement.
- Data governance and market power: policy to prevent monopolization of training data and to enable competitive entry (data trusts, access mandates).
- Regulation, fairness, and institutions
- Alignment and accountability as economic constraints: imperfect alignment and fairness issues affect market outcomes (liability, trust, adoption rates); regulatory standards will shape firm incentives.
- Information and transparency: require disclosure of AI use in economic transactions (labor, credit, education) to enable markets and regulators to function.
- Sectoral implications
- Education and human capital formation: AI in education may alter returns to traditional schooling; need for evidence on learning outcomes and inequality.
- Health and productivity: AI can raise productivity in diagnosis/treatment but raises access, liability, and equity issues that affect welfare distribution.
- Creative industries: co-intelligence changes IP, labor markets for creators, and valuation of originality — with distributional and market-design consequences.
- Policy priorities
- Invest in measurement infrastructure and public datasets to study AI impacts.
- Design labor-market policies that combine retraining, portability of skills, and income supports.
- Update competition and data policies to limit concentration of AI power and capture of rents.
- Institutionalize multidisciplinary governance—ethics, economics, and technical oversight—to steer co-intelligence toward broadly shared benefits.
- Practical implications for economists and policymakers
- Reframe empirical questions around hybrid human-AI productivity rather than AI-as-replacement only.
- Anticipate fast-moving equilibria: policy must be adaptive and evidence-driven, using rapid-cycle evaluation.
- Recognize normative choices: many economic outcomes will hinge on political decisions about distribution, not only on technology.
Overall, the book is a conceptual road map for understanding how AI-mediated co-intelligence reshapes economic relationships, and it calls for an empirical and policy program that treats outcomes as contingent on institutional choices rather than technological determinism.
Assessment
Claims (9)
| Claim | Direction | Confidence | Outcome | Details |
|---|---|---|---|---|
| There were now a hundred million ChatGPT users in two months. Adoption Rate | positive | high | number of ChatGPT users |
n=100000000
100 million in two months
0.12
|
| Less than a year after its debut, hundreds of millions of individuals on all seven continents were using large language models, in virtually every field of professional activity, and in most languages. Adoption Rate | positive | high | number and breadth of large language model users across professions and languages |
hundreds of millions (global adoption across fields and languages)
0.06
|
| The internet had to cope with more or less a decade before it could reach one billion users; social media did it in half times. Adoption Rate | positive | high | time-to-reach one billion users for internet and social media |
internet: ~1 billion users in ~10 years; social media: ~1 billion users in ~5 years (author's phrasing)
0.06
|
| Artificial intelligence has become a partner in our everyday activities: it dictates our emails, diagnoses our diseases, educates our young children, controls our budgets, creates our artworks, and influences the policies made by governments and corporations. Adoption Rate | positive | high | presence/role of AI across a range of everyday activities (email composition, medical diagnosis, education, budgeting, art creation, policy influence) |
0.06
|
| None of the past technologies have spread into so many aspects of human life, so fast. Adoption Rate | positive | high | relative speed and breadth of technological diffusion |
0.02
|
| The concept of co-intelligence describes a new cognitive ecology where the human and artificial minds mutually influence one another to come up with ways of comprehending, creating and making choices that neither of them could accomplish individually. Innovation Output | positive | high | emergence of novel joint human-AI outputs/decisions |
0.06
|
| The opportunities of AI in human good are real and vast; and the opportunities in human ill, in human society, in human institutions of government, and in the longer term in the environment in which humanity thrives are real and underestimated. Consumer Welfare | mixed | high | magnitude of benefits and harms from AI across society, governance, and environment |
0.02
|
| The machines are increasingly becoming competent. Automation Exposure | positive | high | AI capability/competence over time |
0.06
|
| AI will have social, economic, and political impacts on work, inequality, democracy and power. Employment | mixed | high | impacts of AI on employment (work), inequality, democratic processes and power distribution |
0.02
|