Humans must retain control of AI's development

Editor's note: The Government Work Report released last week highlighted the goal of developing the smart economy. National Business Daily spoke to Li Meng, former vice-minister of science and technology, on the potential of the smart economy and the development of artificial intelligence. Below are excerpts of the interview. The views don't necessarily represent those of China Daily.

Although intelligent technologies' large-scale entry into the consumer market may still take time, intelligent agents capable of delivering emotional value are likely to be among the earliest artificial intelligence applications adopted in households.

Companion robots may become the first specialized embodied intelligent systems to enter the household consumer market. At the same time, smart toys represent a potentially massive yet often overlooked industry. These toys may incorporate weak AI or embodied intelligence, integrating functions such as companionship, education and childcare.

As global demand rises, the size of the companion robot market could eventually rival that of the traditional toy industry. Compared with normal robots, smart toys have relatively simple structures. They do not require humanoid forms, dexterous hands or complex multi-joint designs because their core is natural language interaction capabilities. By contrast, household service robots carry high market expectations but face far greater technological challenges.

The smart economy can be viewed as an upgraded version of the Internet Plus economy. Its development will continue to rely on the existing internet ecosystem and technological foundations, while gradually building a new ecosystem centered on networks of intelligent agents.

Policymakers should carefully manage the transition between concentrated job displacement and the distributed creation of new employment opportunities brought about by AI, establishing transition mechanisms and safety nets to smooth this adjustment period.

Ultimately, the relationship between AI and humanity should be defined as that between humans and a super tool, or a tool-like partner.

AI capabilities may progress along a trajectory from assistant, to agent, to embodied humanoid systems, then to a stage where carbon-based and silicon-based intelligence mirror each other, and then eventually toward super intelligence.

Without appropriate constraints and governance, it is conceivable that AI could evolve into partners that resemble humans or even surpass human capabilities, drawing on advances in biotechnology, information technology, materials science, precision engineering and brain science to create human-like or superhuman silicon-based intelligence.

The key question, however, is whether humanity needs or desires AI to develop in this direction, and whether humans have the ability to guide AI's evolution toward a form consistent with human intentions.

The appropriate development direction is for AI to remain a tool that is capable of enhancing human abilities and efficiency, or at most a tool-like partner that provides assistance and emotional value.

For enterprise applications, AI will primarily evolve into super tools that enhance productivity. For consumer applications, it will mainly function as tool-like companions. In either case, it should not develop beyond the boundaries of a tool.

Human beings must remain the central actors in the development of AI and maintain a leading role in shaping its future. Humanity should define the trajectory of AI.

Achieving this goal will require agile governance and strengthened international cooperation so that AI can become an international public good that benefits all of humanity.