Our capacity to generate knowledge and intervene in the world is expanding faster than our ability to understand the consequences. Artificial intelligence (AI) can detect patterns across vast datasets, assist scientific discovery, and model complex processes. Yet the latest planetary boundaries assessment finds that seven of nine boundaries have been crossed. Climate, biodiversity, freshwater, land use, and ocean chemistry are changing through interacting pressures that no single discipline, institution, or technology can adequately address. The contradiction is striking: As our instruments become more capable, the systems on which life depends are becoming less stable.
This is the challenge at the heart of hybrid intelligence. Natural and artificial intelligences bring different strengths to the task of understanding the world. Humans contribute embodied experience, relationships, values, judgment, and responsibility; AI contributes speed, scale, computational synthesis, and the ability to reveal patterns beyond ordinary perception. Their complementarity can enlarge what we collectively know and personally do. But the quality of this relationship depends on whether people retain the capacity and volition to direct it. Technological power can grow while the agency required to use it wisely diminishes.
The planet as teacher
The proposal for natural general intelligence offers a useful way to rethink this relationship. Its starting point is that the Earth is a vast, interconnected system whose dynamics cannot be learned from human-generated text alone. A nature model would draw directly on observations of oceans, atmosphere, ice, land, and living ecosystems; use them to anticipate possible consequences; and learn again from the Earth’s response. Observation, modelling, intervention, and feedback would form a continuous cycle. The planet is a teacher whose lessons cannot be recovered from a database once the moment for observing them has passed.
There is a larger lesson here. Human activity has long altered ecological systems through chains of consequences we barely perceived: Fertilizer applied to a field can transform a distant waterway; land clearing can change rainfall; shifts in ocean circulation can reshape harvests far away. Better models could make these connections more visible and help us choose interventions with greater care. Yet the model cannot settle what ought to be protected, whose needs should count, or which risks are legitimate. Those judgments arise from human and collective agency, informed by science and exercised through accountable institutions.
The fragile core of hybrid intelligence
Agency involves more than having a human approve an AI-generated recommendation. It is the lived capacity to question a premise, interpret evidence, imagine alternatives, choose deliberately, and accept responsibility. These capacities can grow through interaction with AI when technology helps us see more, learn faster, and test our assumptions. They can also weaken through habitual delegation, particularly when convenience rewards acceptance over inquiry. Research on critical thinking among AI-assisted knowledge workers associates higher confidence in the tool with lower reported critical-thinking effort. An experiment on creativity found that AI-assisted stories were rated more favorably while becoming more similar to one another. These findings do not establish inevitable cognitive decline; they expose a design choice with potentially far-reaching social consequences.
The danger is cumulative. If individuals become less practiced in independent judgment, institutions may grow comfortable with automated decisions that few can explain. As those systems spread, their recommendations can shape what is seen as normal, feasible, or desirable. The resulting dependence may make it harder to contest the priorities embedded in technology. A society can become more efficient at executing decisions while losing some of its capacity to deliberate about the direction of travel. For planetary health, where difficult choices involve competing interests, uncertainty, and longtime horizons, that loss would be especially costly.
The hybrid tipping zone
The hybrid tipping zone describes the interaction of four levels of change. At the micro level, agency decay affects how individuals perceive, judge, and act. At the meso level, AI mainstreaming reshapes work, education, communication, and institutions. At the macro level, the race for AI supremacy influences investment, governance, and the concentration of power. At the meta level, planetary deterioration changes the material conditions under which all three unfold. Agency decay is indirectly both the cause and consequence of these wider developments. Their effects accumulate through feedback, creating the possibility of reinforcing trajectories in either direction.
A negative spiral is easy to imagine: Weaker agency encourages uncritical adoption; dependence and competition reward systems optimized for speed, scale, and influence; their social and environmental costs become harder to challenge; worsening uncertainty creates further demand for automated answers. A positive spiral is equally conceivable, although it requires deliberate cultivation. AI can strengthen people’s understanding of complex relationships, support more participatory decisions, and make ecological consequences visible earlier. Better decisions can build trust and institutional capacity, creating conditions in which human agency develops further. A tipping point here is a potential shift in system dynamics, not a claim that one measurable threshold has already been reached.
Learning to flourish within limits
Planetary boundaries research reminds us that the Earth’s systems interact across scales, often in nonlinear ways. A gain in one sector can transfer costs elsewhere; technological efficiency may reduce resource use per task while encouraging greater total consumption. AI itself depends on electricity, cooling water, minerals, land, and industrial supply chains. The energy and material demands of AI must be assessed alongside its contributions to scientific discovery, ecological monitoring and resource management. A credible account of progress needs to ask whether a technological intervention improves the conditions for life overall, including who receives its benefits and who bears its burdens.
Artificial Intelligence Essential Reads
This calls for double literacy. Human literacy is a holistic understanding of self and society, people and planet, including the relationships between inner states, social conditions, collective structures, and ecological systems. Algorithmic literacy means understanding how technology shapes what we notice, how we interact, and the choices we make. Together they equip people to engage with AI as thoughtful participants rather than passive recipients. They also offer a practical foundation for ProSocial AI, which evaluates technological development through purpose, people, prosperity, and planet.
Hybrid flourishing begins from the inside out and is nurtured from the outside in. People need the mindset and agency to act with awareness; they also need institutions, technologies, and ecological conditions that make such action possible. Natural general intelligence could help AI become more responsive to the planet. ProSocial AI asks that the relationship between humans and AI strengthen the capacity to choose, cooperate, and care for that planet. Hybrid intelligence is the cause and consequence of this evolving complementarity. Its future will be decided as much by the human capacities we cultivate as by the machines we build. The question is whether growing AI will enlarge our shared ability to live well within a living world—and whether we will remain capable of deciding what living well means.