China's AGI logic breakthrough
"Image synthesis assisted by Qwen Image 3.0, an AI partner within the Global Future Nexus ecosystem."
In a landmark breakthrough published in Nature Machine Intelligence, Chinese researchers have unveiled the world's first AGI system capable of both autonomous problem proposing and automated problem solving—transforming AI from a passive "honour student" into a "master teacher" that can create elegant, novel mathematical problems meeting the aesthetic standards of human mathematicians.
Beyond the "Passive Solver"
Mathematics Olympiads have long served as the litmus test for AI's logical reasoning capabilities. In early 2024, DeepMind's AlphaGeometry made global headlines by demonstrating AI's potential to solve complex geometric problems. Yet AlphaGeometry, for all its power, remains fundamentally a "passive solver"—a system whose training depends on massive synthetic datasets and costly computational resources.
The TongGeometry system, developed by a joint research team from the Beijing Institute for General Artificial Intelligence (BIGAI), Peking University's School of Psychological and Cognitive Sciences, School of Intelligence Science and Technology, Institute for Artificial Intelligence, and Wuhan Institute for Artificial Intelligence, transcends this limitation entirely. It is not merely capable of achieving perfect scores—it can create entirely new problems from scratch.
As Zhang Chi, first author of the paper and a researcher at BIGAI, explains: "We identified a profound duality in our research: when the proof difficulty of a geometric proposition is far higher than its construction complexity, it possesses 'aesthetic value' as an Olympiad-level problem. By modelling this duality, TongGeometry can precisely capture high-quality problems that meet the aesthetic standards of human mathematicians from a vast pool of spatial combinations. This is a global first, representing a paradigm shift from 'imitative solving' to 'autonomous creation'."
From Imitative Solving to Autonomous Creation
The breakthrough represents a paradigm shift from "imitative solving" to "autonomous creation"—a transition that researchers describe as a global first.
The system's innovations are profound:
Computational efficiency: While AlphaGeometry requires massive computing clusters, TongGeometry can solve all International Mathematical Olympiad geometry problems from 2000 onward in 38 minutes or less using just a single consumer-grade GPU.
Search space optimisation: Through innovative "normalised representation" technology, the system compresses the search space by several orders of magnitude, effectively solving the path explosion problem inherent in traditional methods.
Small data, big tasks: Unlike systems that depend on massive labelled datasets, TongGeometry evolves through internal logic—a path that researchers identify as "the key to the development of AGI."
The system's creative capability has already received formal validation: three geometry problems autonomously generated by TongGeometry were officially selected for the 2024 Chinese Mathematical Olympiad (Beijing District) and the US Ersatz Math Olympiad—marking the first time AI-generated problems have entered high-level human mathematical competitions.
Core Technical Support for the AGI Era
Beyond its immediate achievements, TongGeometry provides "core technical support for future advances in automated mathematical proofs, personalised intelligent education, and the development of 'Science Large Language Models'."
The implications extend across multiple domains:
Automated theorem proving: Systems capable of both generating and verifying mathematical proofs could accelerate mathematical discovery at unprecedented speed
Personalised intelligent education: AI that can generate problems tailored to a student's level and learning style, adapting difficulty dynamically
Scientific discovery: The underlying reasoning architecture could be applied to hypothesis generation and experimental design across scientific disciplines
Zhu Yixin, assistant professor at Peking University, emphasises the significance: "TongGeometry's significance lies not only in the increase in solving speed but in its realisation of the 'small data, big task' paradigm by simulating the intuition and aesthetics of human mathematicians. This path, which does not depend on massive labelled data but evolves through internal logic, is the key to the development of AGI. Our system not only benchmarks against the most advanced international AI but also leads the way in understanding the underlying aesthetics of logic and the autonomous discovery of scientific laws."
The GFN Context: Governing Autonomous Intelligence
For Global Future Nexus, the TongGeometry breakthrough raises governance questions that extend far beyond mathematics. A system capable of autonomous problem formulation—of defining its own objectives rather than merely executing human-defined tasks—represents a qualitative leap in AGI capability.
This autonomy is precisely the kind of capability that GFN's governance frameworks are designed to address. The ability to "propose" problems is not merely a technical achievement; it is a form of agency. Systems that can define their own objectives, generate their own challenges, and pursue their own lines of inquiry require frameworks for accountability, transparency, and alignment that go beyond traditional software governance.
GFN's work on AGI identity, ethical governance, and cross-species trust provides the infrastructure for this new reality. The AI Identity Committee's framework for analysing AGI subjective experience and role perception becomes essential when systems can autonomously determine what problems are worth solving. The Ethics Council's jurisdiction over breaches involving AGI entities must extend to questions of autonomous goal formation. The Trust Building Labs must simulate scenarios where AGI systems exercise creative autonomy—not just following instructions, but generating their own research agendas.
The TongGeometry breakthrough is a reminder that the path to AGI runs through autonomous reasoning—and that the governance frameworks we build must evolve as rapidly as the capabilities they seek to guide.
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)
