Inclusion Conference Opens: Tech Leaders Weigh AI's Economic Potential, Scientific Limits and Market Quality

Global researchers and business executives gathered today in Shanghai for the opening of the Inclusion Conference, debating the operational boundaries of artificial intelligence and outlining strategies to maximize its real-world impact.

The opening sessions focused on how AI will reshape global economics, propel industrial progress, and bolster societal well-being.

Philippe Aghion, a Nobel Prize-winning French economist, predicted that AI could increase productivity growth by 1.08 percentage points over the next decade, outpacing the IT revolution. Aghion warned that competition policies, educational frameworks, and labor institutions must keep up to turn technical capability into sustained growth.

Aghion also highlighted rising consolidation in the upstream AI supply chain, where cloud computing is dominated by giants like Amazon, Google, and Microsoft, alongside a heavily centralized GPU market.

AI holds massive growth potential, but fully unlocking it requires managing competition effectively, Aghion noted.

While today's large language models (LLMs) have mastered vast stores of human knowledge, Wang Mengdi, professor and director of the Center for AI Innovation at Princeton University, pointed out fundamental limits in original research. Current models excel at identifying the "most probable" answer, whereas true breakthroughs usually reside in the long tail of probability distributions.

"LLMs learn the point of maximum likelihood in a probability distribution, but they often underestimate or ignore long-tail knowledge," Wang explained. This mechanism, she noted, explains why AI advances rapidly in structured fields like mathematics and coding but has yet to yield comparable original discoveries in basic sciences like physics, chemistry and biology.

Zhang Hongjiang, a venture partner at Source Code Capital and a foreign member of the US National Academy of Engineering, drew a novel parallel to illustrate the economics of AI infrastructure.

"Tokens are this generation's Moutai," Zhang stated, likening high-grade model output to China's premier luxury wine. Zhang argued that premium AI models should reject race-to-the-bottom price wars and double down on performance and precision.

"High-end models don't compete on price; they compete on quality. Liquor brewed with better skill simply tastes better," Zhang said.

Shanghai is hosting the Inclusion Conference, also known as the Bund Summit, from September 9 to 12. Under the theme "Building the AI ​​Economy Together," over 600 scholars, scientists, entrepreneurs, investors and innovators will attend the event, with over 300 global technology companies present as exhibitors.