In recent discussions regarding the evolution of artificial intelligence (AI), a notable comment came from Zhou Jian, the founder and CEO of Lanma Technology, during the 2024 China Enterprise Competitiveness Annual ConferenceHe claimed, "With the emergence of ChatGPT, it’s evident that artificial intelligence technologies are beginning to generalizeFrom our perspective, AI Agents represent a new form of productive force." Zhou's insights unveiled a transformative moment in the arena of AI, suggesting that the capabilities of such technologies are now solidifying into a new paradigm.

The forum, held from June 12 to 13 by the China Economic Media Think Tank, included discussions surrounding "The Leap Forward of Chinese Business Over the Last 20 Years and the Transformation Towards a New Business Civilization." Zhou's voice resonated throughout the event, providing a multi-dimensional view of the current landscape of artificial intelligence and its implications for business transformation.

In the past two years, market trends have witnessed a remarkable surge in large-scale models, marking the dawn of a new phase in AI technology development

Zhou noted that these large models have opened up new opportunities by fundamentally altering the paradigms of human-machine interactionHe discussed the transition from mere human-machine collaboration to a more integrated human-machine fusion, projecting potentially a future where AI agents and humans exist symbiotically"We may just be witnessing a breakthrough point, with rapid advancements on the horizon," he remarked.

Reflecting on the evolution of AI, Zhou emphasized that this is not a sudden phenomenon but rather a culmination starting from the era of big data, cloud computing, and the groundbreaking AlphaGo event which marked the introduction of AI 1.0 back in 2015. The rise of ChatGPT signifies a clear trajectory toward the generalization of AI technologiesZhou stated that traditional software paradigms largely focused on information systems and data recording, whereas large models introduce a knowledge system that allows software to mimic human-like processes

• Ulta Beauty Beats Q3 Earnings, Shares Surge After Hours
• Bitcoin Surge Continues
• SpaceX's Starlink Constellation Expands
• U.S. Stock Market Appears Stable
• AI Agents: A New Force in Productivity

This shift notes a significant evolution in the digital landscape equivalent to a modern industrial revolution.

What does it mean to be an AI Agent today? Zhou conceptualized the AI Agent as a new form of productive force, enhancing operational capabilitiesHe explained, "While large models provide foundational capabilities, for an AI Agent to operate as an expert, it needs a robust domain modelDespite current cost barriers associated with large models, adopting engineering approaches can potentially minimize these costs and enhance accuracy." In summary, Zhou posited that an AI Agent comprises expert knowledge, models, data, and computing power—four essential factors for effective productivity.

The popularity of ChatGPT has prompted diverse sectors to regard large model technology as a pivotal driver for digital transformationZhou pointed out that the digital metamorphosis has evolved from relying on IT and big data to now emphasizing large model integrations—where AI increasingly assumes central roles in organizational operations

He highlighted existing corporations like Didi and Meituan, which utilize AI as an organizational brain to efficiently allocate tasks among drivers and delivery personnelZhou articulated a vision where declining costs would enable companies to harness AI in a cost-effective manner, potentially reshaping business models and organizational structures.

As Zhou elaborated, "Currently, the ratio of human employees to AI agents could be as steep as 100:1 or even 1000:1; however, in the future, this could invert to 1:10, 1:100, or even 1:1,000,000, a phenomenon we might refer to as super individuality." This intriguing prospect hints at a transformation that could redefine workforce dynamics and operational efficiency.

Nevertheless, Zhou acknowledged that applying large models practically still faces challenges, particularly concerning the necessity for expert knowledge as a precursor for successful implementation

He remarked, "Expert knowledge creates a ceiling for AI Agents." The complexity and specialization required in advanced expert knowledge can hinder universal applicability, underscoring the fact that identical data and models can yield different value based on the expertise infused into the system.

This led him to underscore the importance of empowering AI Agents with expert knowledgeIn doing so, organizations can enhance the productivity of frontline employees while simultaneously capturing a wealth of expert insights embedded within business processesUltimately, this synthesis could provide AI agents with the cognitive abilities necessary for deeper understanding and create robust large action models for enterprises.

Moving forward, how should companies approach the transition to generative AI? Zhou outlined that many large organizations are already investing in this transformation by allocating budgets for computational power, fostering an environment conducive to learning through tools like GitHub Copilot, which enables engineers to generate code using large models

He noted that administrative staff can leverage AI tools like WPS AI for creating diverse document templates, resulting in significant efficiencies across various operations.

Next comes the imperative of private deployment—organizing internal information through AI for streamlined accessZhou proposed organizing internal documents into an AI-retrievable format, establishing a knowledge center that allows employees quick access to crucial resources such as standard operating procedures (SOPs), product manuals, pricing documents, and historical meeting records.

The subsequent phase involves defining the most suitable customer scenarios for large model applicationsZhou pointed out that AI Agents can infuse expert intelligence into existing processes, rendering previously impractical tasks feasibleFor example, in the insurance sector, clients may upload medical examination reports online, prompting insurance brokers to reach out with tailored recommendations

AI could enhance this by utilizing a curated pool of knowledge to generate personalized insurance product suggestions and marketing language, leading to higher conversion rates in policy sales.

But why have traditional entities struggled to achieve such efficiencies in the past? According to Zhou, the main hindrance was the sluggish speed of information transmissionEstablishing intermediate management layers was necessary to counteract this issueToday’s AI Agents, capable of comprehending, transmitting, and summarizing information, hold the potential to fundamentally alter organizational structures and business methodologies.

Looking ahead, Zhou predicts a trajectory where the capacities of large models evolve, transitioning from human-machine collaboration to fusion, and ultimately to coexistenceHe emphasized that while every technology may reach a saturation point, we currently seem to be on the precipice of rapid advancements

His expectation is optimistic; when these technological innovations peak, they may unlock new avenues for human-machine coexistence, albeit he acknowledges that not every process will transform into a fully-fledged industrial brain capability.

“Thus, we envision AI evolving through three stagesThe current phase focuses on enhancing various roles through large model capabilities to automate and augment job functionsThe next stage anticipates AI possessing emotional understanding and interaction ability, eventually managing workflows in a human-machine collaborative organizationThree to five years from now, it is Lanma Technology's vision that individuals will be enabled to design their own intelligent agents,” Zhou articulated.

However, he cautioned against unbridled optimism regarding Artificial General Intelligence (AGI). The development of AGI will require stringent quality control