On November 12, a significant event unfolded at the Baidu World 2024 Conference, where its founder and chairman, Li Yanhong, expressed a powerful prediction about the impending explosion of artificial intelligence (AI) applicationsReferring to this period as the "moment of brilliance for AI applications," he underscored the transformative potential of AI in reshaping economic and social structures.

During his keynote speech, Li highlighted the emergence of intelligent agents as the most mainstream form of AI application, which he believes is on the brink of a significant breakthroughA major highlight of the conference was the unveiling of a no-code tool named "Miao Da," designed to empower users to bring their ideas to life without any programming skillsThis user-friendly software is set for a launch in the first quarter of the following yearLi articulated the essence of AI as an unprecedented opportunity for industrial revolution, emphasizing its capability to expand productivity limits.

Another eagerly anticipated product showcased was the "Little Du AI Glasses," touted as the world's first native AI glasses equipped with a Chinese large model

These glasses are scheduled for release in the second half of the coming year, further illustrating Baidu's commitment to integrating AI into everyday life.

Throughout the past two years, the discussion surrounding large models has become increasingly vibrant, gaining traction across various sectorsMajor companies and startups alike are racing to tap into the potential of generative AI, moving quickly to create applications that utilize these advanced modelsOne critical issue has been the "hallucination problem," the tendency of large models to generate misleading or incorrect outputs, which many consider a significant barrier to broad application.

To tackle this challenge, Baidu has been developing a technology known as Retrieval-Augmented Generation (RAG). Li noted that RAG has transitioned from Baidu's unique approach into an industry-wide consensus, effectively reducing hallucinations in text-based outputs

• Gold Soars: $3000 in Sight for Next Year?
• Quantum Computing Poses Risk to Cryptocurrencies' Stability
• Climate Rules Hit Traditional Energy Firms
• PE Firms Push for Balanced Portfolios
• East vs. West: China's Golden Dragon Soars as US Stocks Dip

While significant progress has been made on the textual front, he acknowledged that further integration of RAG into multi-modal applications, such as image generation, remains an area for improvement.

In this context, Baidu unveiled its image-based RAG technology, iRAG, which merges its vast image database with robust foundational model capabilities to produce highly realistic imagesDuring the presentation, Li demonstrated an image generated by the Wenxin large model, featuring a popular car model soaring over the Great WallThis output showcased the technology's ability to accurately render details without errors commonly associated with hallucinations, marking a substantial step forward in the utility of AI-generated imagery.

iRAG stands to revolutionize various creative domains, including film, comics, storyboards, and poster design, significantly cutting down production costs while enhancing quality

Li reiterated the importance of eliminating hallucinations among large models as a foundational step for the upcoming explosion in AI applications.

As testament to the rising demand for AI services, Baidu reported a staggering daily usage rate of over 1.5 billion calls for its Wenxin large model by early November, a figure that has skyrocketed from 200 million just six months prior—a testament to the intense growth trajectory in China's AI industry.

Centrally, Li Yanhong envisions intelligent agents taking precedence as the pivotal force driving AI applications into the futureHe characterized these agents as fundamental to the AI-native era, similar to how websites dominated the PC age and social media accounts became key in the mobile eraIntelligent agents, according to Li, will emerge as the new conduits for content, services, and information.

By breaking down the various types of intelligent agents into four main categories—companies, roles, tools, and industries—Li illustrated the vast potential in this space

Notably, Baidu has placed a significant emphasis on developing a wide array of intelligent agents, already attracting 150,000 enterprises and 800,000 developers to its Wenxin agent platform.

Li provided a concrete example of how company-focused intelligent agents can transform user interactionTraditionally, businesses operated static websites to showcase their services, but with intelligent agents, they can now provide customized product recommendations and immediate customer support—effectively enhancing engagement and conversion ratesAs demonstrated, the official intelligent agent for BYD Motors led to a stunning 119% increase in sales conversion rates.

Moving on to the no-code tool "Miao Da," which features capabilities for no-code programming, intelligent agent collaboration, and multi-tool usage, Li heralded the dawn of an era where creativity could translate directly into profitable ventures without extensive technical expertise.

The presentation also delved into Baidu's multifaceted approach to redefining the Wenxin large model ecosystem, incorporating innovations aimed at meeting the diverse demands of both consumers and businesses

For enterprise users, Baidu's Executive Vice President, Shen Dou, shared insights into how Baidu Smart Cloud is providing tailored solutions to pressing industry needs, encompassing a collaboration with hundreds of state-owned enterprises and numerous private businesses in pushing the envelope on AI innovation.

The Baidu Smart Cloud platform has already facilitated fine-tuning of over 33,000 models and developed more than 770,000 applications, reflecting the increasing synergy between AI technology and practical applications in enterprisesWith the emergence of a new type of AI infrastructure combining enterprise-level model engineering platforms and heterogeneous computing power, Baidu aims to cement its lead in driving large models' application across various productivity scenarios.

As the ecosystem surrounding the Wenxin large model expands, Baidu's Chief Technology Officer, Wang Haifeng, revealed that the user base for Wenxin Yiyan reached an impressive 430 million, with 18.08 million developers utilizing the Fei Zhang neural network framework to serve over 430,000 enterprises and create 1.01 million models.

Beyond software enhancements, Baidu is actively migrating its AI capabilities onto hardware, exemplified by the launch of "Little Du," the AI glasses, designed to incorporate an array of interactive features such as real-time translation and encyclopedic information retrieval