On September 30, 2025, the bustling world of artificial intelligence was abuzz with a fresh round of predictions and debate, as Zhang Peng, CEO of China’s AI powerhouse Zhipu AI, weighed in on the elusive concept of artificial superintelligence (ASI). While some industry leaders have been raising eyebrows with bold forecasts, Zhang took a more measured stance. According to Reuters and TechNode, he argued that the arrival of true ASI by 2030 is "impossible," at least in the sense of a machine that can fully outthink and outperform humans across the board.
"People reach different conclusions when discussing this issue. I think achieving or exceeding human intelligence levels by 2030 might mean surpassing humans in one or several aspects, but likely still falling far short in many areas," Zhang said, as quoted by TechNode during the launch of Zhipu AI’s latest large language model, GLM-4.6. His words echoed through a community that has been both excited and anxious about the speed of AI progress.
This tempered view stands in stark contrast to the optimism of other tech titans. Sam Altman, the high-profile CEO of OpenAI, recently predicted that ASI could arrive by the end of this decade. On a similar note, SoftBank Group’s Masayoshi Son has speculated that such technology could be a reality as soon as 2035. These divergent opinions highlight just how tricky it is to pin down what "superintelligence" really means—and when, if ever, it might materialize.
So, what exactly is ASI? In the simplest terms, it’s an artificial intelligence that not only matches but surpasses human capabilities in every possible domain—reasoning, creativity, emotional intelligence, and beyond. For years, the concept has been a mainstay of science fiction and, more recently, a subject of heated debate among researchers, ethicists, and policymakers. But as Zhang pointed out, the very definition of ASI is slippery. "The concept is too unclear to outline the specific timeline for superintelligence," he said, casting doubt on the feasibility of setting any concrete date for its arrival.
Despite the uncertainty, the race to build ever more powerful AI continues at a breakneck pace. Zhipu AI, founded in 2019, has quickly risen to become one of China’s most formidable AI firms. In fact, OpenAI itself cited Zhipu AI in June as one of its fastest-growing competitors, underscoring China’s aggressive push to expand its AI footprint both at home and abroad.
The company’s latest creation, GLM-4.6, was unveiled on the very day of Zhang’s remarks. This new large language model comes packed with enhanced capabilities in reasoning, writing, coding, and agent applications. According to TechNode, GLM-4.6 represents a significant leap forward for Zhipu AI, aiming to challenge the dominance of Western-developed models like OpenAI’s GPT-4. The model’s release is seen not just as a technical achievement, but as a statement of China’s growing ambitions in the global AI arena.
But even as these models grow ever more sophisticated, Zhang’s cautionary tone serves as a reminder that there’s a gulf between impressive machine performance in narrow tasks and the kind of general, adaptable intelligence that characterizes human thought. "Surpass in part, fall short in many," he summarized, neatly capturing the current state of the field.
It’s not hard to see why this debate has become so charged. On one hand, the recent advances in AI—particularly in natural language processing, image recognition, and autonomous systems—have been nothing short of astonishing. Just a decade ago, the idea that a computer could write essays, translate languages, or generate realistic images on demand seemed far-fetched. Now, such feats are routine, and the promise of even more powerful models is tantalizingly close.
On the other hand, skeptics like Zhang argue that these successes mask deeper challenges. True intelligence, they say, is about more than pattern-matching or statistical prediction. It involves understanding context, reasoning through ambiguity, and adapting to novel situations—areas where even the best AI systems still struggle. The road from today’s specialized, data-hungry models to a truly general intelligence is, in Zhang’s view, much longer and more uncertain than some would have us believe.
Of course, the stakes are high. If and when ASI does arrive, it could upend nearly every aspect of society, from the economy and the labor market to national security and the very nature of human identity. That’s why the question of "when"—and "if"—matters so much, not just to technologists but to policymakers, business leaders, and the public at large.
For now, the field is marked by a mix of optimism, skepticism, and a healthy dose of humility. As Zhang put it, "People reach different conclusions when discussing this issue." And it’s not just a matter of technical capability; it’s also about values, ethics, and the kind of future we want to build. Some see the rapid advance of AI as a cause for celebration—a chance to solve big problems and unlock new possibilities. Others worry about the risks of unchecked automation, bias, and loss of control.
Interestingly, the debate over timelines and definitions is nothing new. Decades ago, pioneers like Alan Turing speculated about the possibility of machines that could "think" like humans. Yet, despite waves of progress and hype, the goalposts have kept moving. Each breakthrough seems to reveal new layers of complexity, reminding us that intelligence—whether biological or artificial—is a slippery, multifaceted phenomenon.
As 2025 draws to a close, the release of GLM-4.6 and the ongoing debate over ASI serve as a microcosm of the broader tensions in the AI community. There’s excitement about what’s already been achieved, caution about what remains undone, and a persistent curiosity about what comes next. Will machines ever truly rival the human mind in all its richness? Or will they, as Zhang suggests, always "fall short in many" ways, no matter how advanced they become?
For now, at least, one thing is clear: the journey toward artificial superintelligence is as much about asking the right questions as it is about finding the right answers. And as the world’s leading minds continue to debate, develop, and dream, the rest of us will be watching—sometimes in awe, sometimes in fear, but always with a sense of wonder at what the future might hold.
