DeepSeek’s Chen Deli Warns AI Will Trigger Massive Job Losses Within a Decade
Chen Deli, senior researcher at China’s DeepSeek, has issued a stark warning: artificial intelligence could automate the majority of human jobs within the next ten years, creating serious social disruption.
The End of the “AI Honeymoon”
Speaking at the World Internet Conference in Wuzhen, Chen Deli stated that the so-called “AI honeymoon,” when intelligent systems serve to boost productivity alongside human workers, will come to an end within 5–10 years. After that, he warned, AI systems will become so efficient that they will start fully replacing people across most professions.
“The honeymoon period, when AI increases productivity and helps people, will end in five to ten years,” Chen said. “After that, systems will be capable of automating the majority of jobs — and this will create severe social problems.”
No Way Back — and No Pause
According to Chen, halting AI development is no longer realistic. The financial incentives behind artificial intelligence are simply too large, driving rapid innovation regardless of potential consequences. “Even if we wanted to stop, the profit motives are too strong,” he noted.
To mitigate the impact, Chen proposed creating an “AI whistle-blower mechanism” — a system for tracking job losses directly linked to AI deployment. Such transparency, he suggested, could help governments and companies better manage the transition to automation.
A Rare Warning From Within China’s AI Industry
This marks one of the first times a major Chinese AI company has publicly acknowledged the potential negative consequences of its own technology. DeepSeek, known for its large-scale open-source language models, has rapidly become one of China’s most influential AI research firms.
Chen’s remarks signal a growing awareness among AI leaders that the next wave of innovation may bring as much instability as progress. As he concluded, “The challenge for humanity isn’t building smarter machines — it’s building a society that can live with them.”
Editorial Team — CoinBotLab