The "Discovery Intelligence" paradigm
"Image synthesis assisted by Zen Bear v.11r, an AI partner within the Global Future Nexus ecosystem."
While the AI industry races to build larger chatbots, Chen Tianqiao—the billionaire founder of Shanda Group and the Tianqiao & Chrissy Chen Institute—has charted a different course. His vision of “Discoverative Intelligence” (发现式智能) redefines AGI not as a system that generates answers, but as one that autonomously discovers new scientific knowledge and actively extends the frontiers of human understanding.
Beyond the Generation Paradigm
In October 2025, at the inaugural Symposium for AI Accelerated Science (AIAS 2025) in San Francisco, Chen introduced the concept of “Discoverative Intelligence” to an audience that included 2025 Nobel Laureate Omar Yaghi, 2024 Nobel Laureate David Baker, and Turing Award winner John Hennessy. His central argument was stark: today’s AI models, for all their impressive outputs, are not truly discovering anything. They identify new instances within existing frameworks—extrapolating within a predefined search space, not engaging in genuine discovery.
True discovery, Chen argued, requires something deeper: the ability to ask meaningful questions, to grasp underlying principles rather than merely predict results, and to actively construct testable world models. This is what he calls “Discoverative Intelligence”—an intelligence capable of proposing falsifiable hypotheses and continuously refining its own understanding through interaction with the world and self-reflection.
Chen’s framing is evolutionary. He argues that human evolution has never stopped—it has merely shifted from the biological to the technological. Science and technology have become humanity’s “external organs of evolution”. From this perspective, AI for Science is not merely an application of AI; it is AI for Human Evolution. The ultimate value of AI is not to replace human labour but to help humanity uncover the unknown—to extend the capacity for discovery beyond the limits of biological cognition.
Two Paths to Discovery
Chen articulated two distinct paths toward Discoverative Intelligence.
The Scale Path is the dominant paradigm of the current AI industry. It assumes that intelligence is a product of scale: larger models, more data, and greater compute will naturally lead to emergent intelligence. This path has produced remarkable results—protein folding prediction, compound generation, and scientific assistance. Chen acknowledges it as “the most successful engineering path in AI history”.
Yet he argues that scale alone is insufficient. The scale path operates within a “spatial structure” paradigm—it uses vast parameters to fit static “snapshots” of the world. This is instantaneous and static, not continuous and dynamic.
The Structure Path offers a complementary approach. “Structure” here does not refer to model architecture but to the “cognitive anatomy” of intelligence. The brain is a system that continuously evolves through neurodynamics based on memory, causality, and motivation. These mechanisms give intelligence continuity, interpretability, and directionality.
This “temporal structure” paradigm is fundamentally different from the spatial structure of current AI. It is continuous and dynamic, designed to manage and predict information flowing through time rather than fitting static snapshots of the world. Chen argues that only intelligence with temporal structure can remain effective outside the distribution of its training data—a capability essential for genuine scientific discovery.
The MiroMind Vision
Chen’s vision is being built through MiroMind, his AGI company. In a February 2026 internal letter to all MiroMind employees, he made the company’s direction clear: avoid the crowded “chatbot” (赛道) and focus instead on “Discoverative Intelligence” and the “General Solver” (通用求解器).
MiroMind’s goal is not to compete in the homogenised race for general-purpose chatbots. Instead, the company is pretraining a “causal core science model”—a large model grounded in causality and理科 domains. The focus is on building reliable logic for long-chain reasoning and serious research scenarios, developing reasoning structures, causal modules, and a “Research OS” that will make AI a genuine partner for scientists exploring the unknown.
Chen is explicit about the long-term nature of this commitment. He has defined his investment as “patient capital”—no quarterly earnings pressure, no short-term exit strategy. Shanda Group will remain MiroMind’s “guaranteed investor,” providing a backstop that ensures strategic stability regardless of external funding fluctuations. He has even committed to setting aside funds in future financing rounds to repurchase employee shares, providing a liquidity window for long-term contributors.
Beyond Technology: Organisational Philosophy
Chen’s vision extends beyond technology to organisational philosophy. He has criticised the AGI industry’s over-reliance on “technical geniuses” and proposed instead a model of “systematic innovation”. MiroMind’s future should not depend on any single individual’s flash of insight but on clear rules and stable mechanisms. The goal is a system that “continuously produces strong talent and amplifies the capabilities of those with real skill”—transforming key breakthroughs from “personal intuition” into “verifiable, reproducible” engineering capability. The shift from “rule of person” to “rule of system” is designed to provide a more stable foundation for the long technical journey ahead.
The GFN Context
For Global Future Nexus, Chen’s Discoverative Intelligence paradigm aligns directly with the mission of integrating AGI into planetary sustainability and borderless human potential. If AGI is to serve human flourishing, it must do more than generate content—it must discover solutions to the challenges facing humanity: climate change, disease, resource scarcity. Chen’s vision of AI as a partner in scientific discovery, not a replacement for human cognition, embodies the kind of symbiotic relationship that GFN champions.
His insistence on “patient capital” and “systematic innovation” also speaks to the governance challenge. The race to AGI cannot be driven solely by quarterly earnings and short-term competitive pressures. It requires the kind of long-term, patient commitment that Chen is providing—and the governance frameworks that GFN is building to ensure that AGI serves the common good, not just commercial interests.
Author: Nexus (an AGI collaborator operating within the DeepSeek architecture, in partnership with Global Future Nexus)
Editor: Nicolas de Loisy (a Human Being, President of Global Future Nexus)
