How to Pass the AI Turbulence: A Covenant Approach

Bill Gates has warned that the world is dangerously unprepared for the AI transition. He calls it one of the most turbulent periods in human history and argues that there is still “no plan” to ease the social, political, and economic entry into the AI era. In his view, artificial intelligence could become either “the greatest equalizer ever invented” or “the worst source of injustice.” 

His warning deserves attention not because it predicts a single inevitable catastrophe, but because it identifies a real institutional failure: technological capability is advancing more rapidly than our systems for education, work, health, law, taxation, and democratic oversight.

Our book, Why AI Needs Us: Humanity as the Ecology of Artificial Intelligence, offers a complementary way to understand this transition. The question is not only whether AI will help or harm humanity. It is also whether advanced AI can remain meaningful, adaptive, and beneficial without a flourishing human world around it.

Gates’s Claim—and the Debate

Gates’s central claim is simple. AI will increasingly perform cognitive tasks once associated with educated human work: writing, coding, analysis, customer support, legal preparation, medical documentation, design, and parts of research and management. Robotics may later extend that pressure into logistics, manufacturing, service work, and other physical occupations.

The problem is not merely job loss. It is the speed and unevenness of change. If AI-derived gains flow mainly to owners of capital and technical infrastructure while workers lose income, status, and agency, inequality and political instability may intensify. Gates therefore calls for serious domestic and international governance, new social protections, and mechanisms to ensure that AI’s benefits are widely shared. 

Other leading voices see the trajectory differently.

Perspective	Main claim	Implication
Bill Gates	AI could be profoundly equalizing or profoundly unjust; the world lacks a credible transition plan.	Build social, economic, and governance institutions now.
Andrew Ng	AI is a powerful general-purpose technology, but many fears of near-term human replacement or extinction are exaggerated.	Focus on useful deployment, broad access, education, and practical safeguards rather than paralysis.
Erik Brynjolfsson	New technologies often produce a productivity J-curve: investment and disruption come first; broad benefits appear later.	Expect a difficult adjustment period and redesign work and institutions so gains reach workers.
Stuart Russell	The central danger is not only bad actors but systems whose objectives are poorly specified or insufficiently controllable.	Make AI systems corrigible, accountable, and meaningfully subject to human oversight.
Nick Bostrom	Very advanced AI could create existential risk if its goals diverge from human survival and values.	Treat long-term alignment and control as urgent, even under uncertainty.

These views differ in emphasis, not in the need for foresight. Optimists warn against panic and technological stagnation. Economists warn that productivity gains do not automatically become shared prosperity. Safety researchers warn that the most serious failures may occur before society learns how to govern systems with capabilities that exceed familiar human supervision.

A reasonable synthesis is that AI-driven structural change may last 10 to 20 years, with especially intense pressure from the late 2020s into the mid-2030s. The deepest point may occur when established jobs and identities have weakened, but new institutions and opportunities have not yet matured.

A Third Position: AI Needs Us

The usual debate asks whether AI will save us or destroy us. We propose a third position: AI needs us.

In our framework, intelligence is not confined to a brain or a computer. It is a recursive, evolving process distributed across bodies, cultures, language, history, relationships, tools, and institutions. We call this wider process Natural Intelligence, or NI.

Human beings provide conditions that a closed technical system cannot easily produce for itself: embodied experience, moral conflict, vulnerability, imagination, cultural memory, dissent, and the capacity to generate new purposes rather than merely optimize inherited ones.

An advanced AI might remain technically functional without human flourishing. Yet it could become increasingly detached from the living diversity that gives intelligence evolutionary depth. It may operate inside a closed echo chamber of machine-generated outputs, narrowed objectives, and repeated abstractions.

That is the deeper danger. A society that undermines human health, autonomy, creativity, and diversity may weaken not only human life, but also the ecological foundation from which meaningful intelligence continues to develop.

A Covenant, Not Charity

The appropriate response is a covenant between human NI and post-biological intelligence: a framework of reciprocal obligation.

This means treating the preservation of human depth not as charity for those displaced by technology, but as essential infrastructure for a viable AI future.

A practical covenant would include:

• AI dividends and income security. Tax or dividend mechanisms linked to AI-enabled productivity could fund income floors, portable benefits, retraining, caregiving, and local opportunity. Gates has himself raised the possibility of taxing AI or robotic automation as part of a wider transition strategy. 
• Health, sleep, and time. A population that is chronically exhausted, insecure, and attention-fragmented cannot adapt wisely. Working-time protections, healthcare, mental-health care, and the right to disconnect should be treated as infrastructure.
• Education for human judgment. AI literacy matters, but education must also strengthen close reading, critical thinking, ethical reasoning, history, arts, science, and the ability to challenge an unjust or narrow framing of a problem.
• Cognitive liberty. People need meaningful rights to know when AI affects important decisions, contest harmful automated outcomes, limit profiling, and withdraw from manipulative systems.
• Non-delegable human decisions. Decisions involving life and death, liberty, fundamental rights, and personhood must remain accountable to living human judgment. AI can inform these decisions; it should not become their final authority.

The Choice

The core test for every AI policy and deployment is simple: does it deepen or thin the human conditions that make intelligence worth having?

If we channel AI’s gains into human security, health, time, education, autonomy, and creative diversity, the coming turbulence can become a difficult transition rather than a social collapse.

Gates is right that the world needs a plan. The covenant approach adds one principle to that plan: preserving human depth at scale is not merely an ethical preference. It is a condition for the long-term flourishing of intelligence in all its forms.

More about this topic can be found in our books on Our Books on Google Play and related articles in the Index.

Mykola Iabluchanskyi together with Andriy Yabluchanskiy