The World Artificial Intelligence Conference (WAIC) concluded in Shanghai, drawing crowds comparable to those seen during the Spring Festival travel rush.

Behind this surge lies a shift in the conference's focus: AI has moved beyond its earlier conceptual phase—akin to concept cars on display—and entered a 'mature stage' where it is as readily purchasable and deployable as vehicles sold at a 4S dealership.

One of the most closely watched consumer-facing (C-end) products at this year's event was Honor's Robot Phone. It goes beyond merely adding a gimbal to enhance photography; more significantly, it endows the smartphone with a 'body'—enabling it to perceive its surroundings through cameras and possess reasoning, execution, and feedback capabilities. For instance, if you walk past an item of interest in a shopping mall without noticing it, your phone can still recognize the product and prompt you about whether to purchase it.

Kevin Kelly, often dubbed the 'Father of the Silicon Valley Spirit' remarked on-site that technology and humanity are co-evolving. He emphasized that now that AI has reached maturity, it must demonstrate its value through tangible outcomes. While many AI systems today can hold conversations, those capable of directly delivering concrete results remain scarce. WPS highlighted precisely this delivery capability of its AI at its exhibition booth.

Baidu's exhibition area left a strong impression: elementary school students took the stage to write programs using natural language—made possible by Baidu's 'Dazi' intelligent agent tool. A scenario that would have been unimaginable just a few years ago has now become reality.

Intelligent agents are set to become deeply integrated into daily life. According to an IDC report, the number of active agents worldwide stood at 28.6 million last year, is projected to reach 79.4 million this year, and is expected to soar to 2.216 billion by 2030—meaning that, by then, there will be roughly one active agent for every four people.

JD.com's booth also defied expectations: its exhibits largely diverged from typical e-commerce attributes. JD's unique advantage lies in its operational authority over the physical world—real-world data accumulated across its end-to-end warehousing and logistics chain, including warehouses, shelves, products, couriers, and the behavioral data of hundreds of millions of consumers. AI models trained on such 'real-world data' demonstrate stronger practical capabilities.

Ant Group has ventured furthest along a differentiated path. As future agents handle tasks like booking tickets, grocery shopping, or bill payments on behalf of humans, who will verify transactions, resolve disputes, and ensure security? Ant's long-standing expertise in payments, risk control, credit assessment, and security is poised to serve as the foundational infrastructure for the AI-driven transaction era. At the front end, Ant deploys Afu, Abao; at the middle layer, it offers AI-powered payment and merchant agents; and at the base, it leverages large models, embodied intelligence, and data security frameworks—seeking to answer a critical question: when intelligent agents conduct business, who serves as the trustworthy cashier?

It is precisely these concrete manifestations of 'mature behavior' that explain why, despite the sweltering heat, attendees flocked to the event in droves. What was showcased at the conference may well represent the near-future reality that will soon surround us all.

【本文部分内容由AI辅助生成，特此声明。The author(s) generated part of the content in this work with the assistance of artificial intelligence (AI), which is hereby declared.】