It is widely acknowledged that human intelligence outperforms machine intelligence, given that the human brain is 10 to 100 times more complex than the most sophisticated artificial intelligence models available today. Yet this gap is rapidly narrowing. Humanity may soon confront a technological singularity, a turning point where AI surpasses human capabilities and triggers profound, unforeseen shifts in human civilization.

Nevertheless, an alternative hypothesis suggests that humans' intellectual advantage over other animals stems primarily from growing up and living in human society. This theory also underpins why many psychologists reject the idea of culture-independent intelligence. Human intelligence is fundamentally shaped by environmental influences and social context.

Determining which hypothesis holds true—or whether both are valid—requires rigorous, objective empirical testing. Recent breakthroughs in biological neuroscience and computational research have made such definitive experiments technically feasible.

A core experimental approach centers on the digital reconstruction of biological brains. To date, the most sophisticated digitally cloned neural system belongs to the fruit fly brain. Full brain mapping unlocks its functional potential: researchers can connect the digital brain clone to customized input and output channels that differ entirely from those of a natural biological brain.

For instance, computer scientists can feed audio signals into visual neurons or treat muscle nerve outputs as terminal readouts, regardless of the fruit fly's original biological functions such as flight. With curated training datasets and computational methods including backpropagation, researchers can train the digital brain to process diverse inputs and generate targeted outputs.

Google's digital fruit fly brain, which replicates the exact neuronal wiring of a real fruit fly brain, has supported extensive experimental validation. If intelligence were an inherent property of physical brain structures, digital replicas would struggle to master unfamiliar tasks. In practice, however, the digital brain demonstrates adaptability and versatility.

In controlled trials, digital fruit flies have successfully played gobang (five-in-a-row), simulated car driving, transcribed text such as "Hello, world", and even simulated slicing doner kebabs. When placed in enriched, task-oriented environments, these digital neural systems consistently outperform their biological counterparts. By contrast, real fruit flies lack environmental incentives for complex cognitive tasks and remain instinct-driven, drawn only to basic survival behaviors such as feeding on rotting fruit.

Brain mapping technology also enables precise modulation of neuronal circuit activities in living biological brains. Optogenetics, recognized by the 2026 Nobel Prize in Physiology or Medicine, allows scientists to precisely regulate specific neural circuits—such as those governing the sleep-wake cycle.

Notably, neural circuits with fixed biological functions may serve entirely different roles in digital systems. For example, sleep-regulating neurons in biological brains have no sleep-related function in digital fruit flies, which require no rest. A neural circuit's function is therefore plastic, defined and reshaped by its training and operational environment.

This argument also applies to advanced AI systems. AI does not possess inherent machine intelligence, as all artificial systems are entirely built, trained and nurtured by human agents.

The recent wave of AI breakthroughs in mathematical reasoning illustrates this point. OpenAI has released AI-derived solutions to long-standing unsolved mathematical puzzles that are critical to advancing modern mathematics. Yet these achievements are not autonomous. Without human-defined research priorities, investor funding, and collaborative teams of mathematicians and computer scientists allocating financial and computational resources, AI models would never independently target and solve such complex academic problems.

Philosopher John Searle's Chinese Room Argument further demonstrates the non-autonomous nature of machine intelligence. In Searle's thought experiment, an individual with no knowledge of Chinese is placed in a closed room. Equipped only with a rule-based symbolic program—analogous to a large language model—the person can process incoming Chinese characters and generate fluent, indistinguishable outputs by following preset mechanical instructions.

Despite producing perfect results, the individual inside the room understands nothing of the language, cannot evaluate output quality, and cannot self-optimize the system without external guidance. This analogy proves that AI systems lack full autonomy and cannot achieve endogenous progress independent of human support.

Physician-scientists developing AI algorithms for precision medication timing and dosage recommendations emphasize that medical AI deployment must be bound by robust multi-layered guardrails. These include ethics committee approval, regulatory authorization, and patient consent, as well as interdisciplinary collaboration among data scientists, clinicians, pharmacists and nursing staff.

While AI can exhibit intelligent behaviors, it never achieves absolute autonomous cognition. Beyond regulating AI itself, humanity must upgrade governance tools to constrain and guide human behaviors in AI development and deployment. Traditional physical supervision mechanisms can be modernized and optimized, just as physical fences are gradually replaced by flexible, efficient GPS-powered virtual fences.

To realize the vision of fair, inclusive global AI governance proposed at the 2026 World Artificial Intelligence Conference and High-Level Meeting on Global AI Governance held in Shanghai, multilateral cooperation on next-generation AI regulatory tools is urgently needed.

Machines possess no intrinsic, quantifiable intelligence comparable to that of living organisms. All artificial intelligence is conditional, shaped by the structural, instructional, regulatory and economic systems in which AI operates. Whether AI is viewed as a transformative opportunity or a potential risk, both its promises and perils ultimately stem from human choices. The responsibility for governing, regulating and steering AI development rests entirely with humanity.