Sohu's Charles Zhang Urges Rationality in the AI Era, Advocates for "Human-C...

Charles Zhang, Founder, Chairman of the Board, and CEO of Sohu, and PhD in Physics. Image courtesy of Sohu.

Charles Zhang, founder, chairman, and chief executive officer of Sohu.com, one of China’s earliest internet pioneers, says the key to survival in the era of artificial intelligence (AI) is not blind pursuit of technology but a return to human connections, rational strategy, and cash-flow discipline.

Speaking at the opening of the 2025 World Internet Conference in Wuzhen, Zhejiang province, Zhang—who holds a PhD in physics—shared candid views on the future of AI, the evolution of China’s internet, and Sohu’s distinctive approach amid an industry-wide rush toward large language models.

“AI should be treated as a tool for efficiency, not the strategic core,” Zhang told reporters in a media interview on Nov. 7. “When everyone is chasing AI, one should return to the essence of social interaction. In business, cash flow and rationality must take precedence over hype and ambition.”

A Veteran’s Perspective on the Internet’s Transformation

Having witnessed the rise and reinvention of China’s internet industry over three decades, Zhang reflected on its evolution—from the PC era to mobile, from text and images to livestreaming, and now to AI-driven content creation.

“The internet has transformed from an information network to infrastructure for the digital age,” he said. “We moved from portals to search, then to algorithmic feeds and social platforms. Now, AI generates content itself.”

In this landscape, Zhang believes China holds a global advantage. The country’s massive user base and competitive environment, he said, have fostered rapid advances in algorithmic recommendation and data processing. “We are catching up very well in AI,” he noted.

But while rivals pour billions into foundational AI model research, Sohu has taken a different route. Zhang confirmed that Sohu is not developing its own large models and has refrained from entering the AI infrastructure race.

“The priority for any company should be survival and sustainability,” he said. “You can’t borrow heavily to invest and accumulate debt—that’s extremely risky. Many companies don’t fail because of poor strategy, but because their cash flow breaks.”

Zhang advised entrepreneurs to “become their own CFOs,” calculating input-output ratios rather than following industry hype. “Opportunities must be evaluated rationally and quantitatively,” he said. “You can’t rely solely on passion or trends.”

Efficiency vs. Humanity: The Physics of AI

Drawing on his background as a physicist, Zhang proposed a dual-principle approach to corporate management: “In the short term, focus on data; in the long term, focus on theory.” Companies, he said, must track metrics but avoid being consumed by them. “Believe in your strategic direction, and stay consistent with your long-term theory.”

To illustrate AI’s role, Zhang compared it to “building blocks” in a skyscraper. Traditional search engines, he said, offer users pre-built houses of information, while AI breaks data into smaller bricks to construct precise answers.

“AI makes information access more modular and efficient,” he said. “But the skyscraper still needs blueprints and a solid foundation. For Sohu, that foundation is people, and the structure is social interaction.”

Zhang warned against overreliance on AI, emphasizing that humans must retain independent thought and memory. “The human brain and AI process information in fundamentally different ways,” he said. “Memory stored in the brain is the foundation of deep thinking. Dependence on AI will erode logical reasoning.”

He proposed a new concept: AI turns every person into a “knowing molecule” rather than an “intellectual.” AI, he said, enables anyone to instantly access vast background knowledge, improving collective communication efficiency. “As a whole, humanity will become smarter,” he said.

However, “knowing” is not “understanding.” Teaching physics, Zhang observed that while AI can flawlessly recite and derive formulas, its reasoning is alien to the human mind. “To truly understand, you must follow human cognitive pathways,” he said. “Relying on AI for all answers will weaken the ability to think—and even harm your health.”

“Say No to AI” in Certain Areas

Zhang also advanced a counter-efficiency philosophy: excessive convenience may reduce quality of life. “Humans need participation and activity,” he said. “The better the hotel, the more primitive it feels—manual curtains and wooden desks preserve that sense of engagement. In the AI era, we must set boundaries and consciously say ‘no’ in some areas to protect our humanity.”

In creative fields, Zhang said AI remains far from capable of understanding humor, emotions, or authentic storytelling. “AI-generated comedy videos are completely unreliable,” he said, laughing. “The future’s core competitiveness will come from human thinking, creativity, and emotional resonance.”

His advice: “If you don’t want to lose to AI, enhance your powerful thinking and learning abilities through content creation. Use the irreplaceable parts of your brain.”

Defamation, Extortion, and Privacy: The Three AI Dangers

Zhang expressed grave concern over AI’s potential to amplify social risks, summarizing them with three words: Defamation, Extortion, and Privacy.

He warned that generative technologies make it “unprecedentedly easy and cheap” to create false information. Deepfake tools, he said, could be exploited for character attacks and blackmail, pushing society toward “a fake world” where verifying truth becomes costly and exhausting.

To tackle such problems, Zhang drew parallels to the early 2000s copyright battles between media outlets and search engines. “Ultimately, it was through lawsuits that frameworks like the Safe Harbor Principle were established,” he said. “The same will happen with AI. You can only solve it through legal action.”

Court rulings, he argued, will define new boundaries for platform responsibility and gradually form a regulatory consensus. “It’s a dynamic process of negotiation—there’s no one-step solution.”

Sohu’s Counterintuitive Strategy: Rebuilding Real Relationships

While much of the tech world is reinventing itself around AI, Sohu’s current focus remains firmly within the consumer internet, particularly media and social video. Zhang sees enduring opportunity in these “traditional” sectors.

Sohu’s flagship feature, the “Following Feed,” exemplifies this people-first philosophy. “It’s not just about content consumption—it’s about social distribution, built around relationships between people,” Zhang said.

In the era of short video and user-generated media, he explained, audiences are shifting from passively consuming information to following trusted creators. “Content is no longer just about what, but about who,” he said.

Sohu’s “Following Feed” and “Timeline” features aim to deepen those personal connections. “On other platforms, it doesn’t matter who posts the content,” he said. “On Sohu, we care about the person behind it.”

Zhang also revealed plans to revive Sohu’s American TV content strategy, after scaling back in previous years due to piracy concerns. The company will increase purchases of licensed U.S. dramas and films to enhance its video ecosystem.

Sohu Video will blend long-form and short-form programming—hosting popular vertical short dramas while producing horizontal, cinematic short series in-house.

This “content plus connection” model, Zhang said, aims to restore authenticity in an era dominated by algorithmic feeds. Instead of competing to become another AI-powered platform, Sohu is attempting something counterintuitive: rebuilding genuine social relationships beneath the algorithmic layer.

“Whether it’s reviving American series or investing in live knowledge broadcasts, our distribution logic centers on people,” he said. “We’re constructing a real social graph, not just a content waterfall.”

For Zhang, Sohu’s guiding principle has remained unchanged since its founding in 1998: building communities through communication. The rise of AI, he argued, should not displace that human core but enhance it.

“The internet began as a tool for human connection,” he said. “AI is just another layer of that evolution. But people—our thoughts, emotions, and interactions—must remain the foundation.”

As the world races toward an AI future, Sohu’s founder continues to preach patience, precision, and perspective. “In the end,” Zhang concluded, “what matters most is not how smart AI becomes, but how wisely we humans choose to use it.”