US losing AI race to China, says former Google CEO
What's the story
Former Google CEO Eric Schmidt has warned that the United States is losing the artificial intelligence (AI) race to China. He said that the gap between the two countries in AI development had narrowed to about six months. Schmidt's assessment is a major shift from his previous view, which had estimated China's lag behind the US at one or two years.
Market dynamics
Affordability could tip the scales in AI race
Schmidt believes that affordability, not just technical superiority, could be the key to success in the AI race.
He warned that if Chinese companies continue to offer capable AI tools for free or at a low cost, they could easily outpace their American counterparts.
This is especially true in developing countries, where consumers may not be able or willing to pay for expensive Western alternatives.
Competitive edge
Chinese developers closing the gap
Chinese developers have been releasing AI models that are either free or much cheaper than their Western counterparts.
The availability of open-weight models has also expanded the choices for businesses looking for alternatives to expensive proprietary services.
If these models offer enough performance for everyday tasks, users may have little incentive to choose a more expensive competitor, creating a commercial challenge for American AI firms.
Business dilemma
Balancing act for American firms
American AI firms now have to balance the cost of developing and running increasingly capable systems with the need to attract customers.
Even a technically superior model could struggle to build a sustainable business if cheaper alternatives meet most users' requirements.
This was highlighted by Schmidt's example of Tencent-developed AI agent on WeChat that can examine medicine packaging and provide personalized guidance.
Export controls
Export restrictions may not be effective
Schmidt also questioned the effectiveness of US restrictions on advanced semiconductor exports in limiting China's AI capabilities.
He speculated that some hardware may have been "pilfered or smuggled" into China, allowing developers to access advanced chips despite controls intended to restrict their availability.
His remarks do not establish the extent of any such activity but highlight a potential loophole in export controls.
Technological progress
AI distillation's role in narrowing the gap
Schmidt also pointed to AI distillation as a factor in narrowing the gap between US and China in AI development.
The technique uses outputs from a more capable model to help train another system, potentially allowing developers to reproduce some of its abilities with a different model architecture or at lower operating costs.
Combined with advances in software, such methods can help companies improve performance without relying exclusively on access to the newest hardware.