Baidu World 2024 Announcements: New iRAG Technology, MiaoDa No Code Tool, AI Smart Glasses and More 

At the Baidu World 2024 conference in Shanghai, Baidu founder and CEO Robin Li unveiled new groundbreaking technologies and shared the company’s vision for the future of artificial intelligence (AI). 

On Nov 12, during the Baidu World 2024 Conference, Robin Li, Baidu Inc co-founder, unveiled two new AI-powered tools, iRAG technology, generating pictures based on retrieval-augmented technology, and "Miaoda", achieving programming without coding. He also presented the Wenxin AI… pic.twitter.com/pRh4Ub77qa

— China Daily (@ChinaDaily) November 12, 2024

Among the most notable announcements were the debut of Baidu’s innovative image generation technology, Image-Based Retrieval-Augmented Generation (iRAG), and the no-code development tool, “MiaoDa.”

In addition, Li highlighted ERNIE’s expanding reach and user base and mentioned the introduction of Xiaodu AI Glasses, which is a contender for Meta’s glasses. Here’s everything you need to know about the Baidu World 2024 announcements:

Every Announcement at Baidu World 2024

Tackling Image Hallucinations with iRAG

One of the most persistent challenges in generative AI has been “hallucinations,” where AI models produce inaccurate or non-existent elements, particularly in images. To combat this, Baidu introduced WenXin iRAG, an advanced image generation technology leveraging retrieval-augmented methods and Baidu’s vast image database. This technology aims to produce highly realistic visuals by integrating accurate references into the generative process.

As Li stated: “Over the past 24 months, the AI industry has seen transformative changes, and addressing hallucinations has been a top priority.” Thus, WenXin iRAG seeks to minimise these hallucinations by integrating Baidu’s extensive library of hundreds of millions of images with advanced retrieval-augmented algorithms. By drawing on a vast database, iRAG can generate hyper-realistic, contextually accurate images, positioning itself as a valuable tool for industries like brand promotion, film production, and creative media.

During the conference, Li showcased a demonstration from iRAG of its ability to create ultra-realistic images, emphasising its potential to reduce costs and improve workflows for content creators. Besides, Li spotlighted the potential of this technology to produce visually accurate content and reduce production costs.

MiaoDa: No-Code Development for Everyone

Further underscoring Baidu World 2024’s theme of democratising AI, Li introduced MiaoDa, a no-code tool designed to empower users to develop complex software applications using natural language. For context, MiaoDa is equipped with features like no-code programming, multi-agent collaboration, and multi-tool integration, enabling users to articulate their software needs in everyday language without any coding expertise.

Li demonstrated MiaoDa’s capabilities by creating an event registration system on stage, showcasing how the tool coordinates multiple AI agents to build applications from scratch. “This is the most complex example of multi-agent collaboration to date,” said Li, pointing out that MiaoDa can identify and fix software bugs, manage projects, and facilitate content editing autonomously. 

The tool’s seamless interaction with Baidu’s other services, such as maps and web search, provides a comprehensive development platform that significantly lowers the barriers to entry for non-technical users.

Agents: The New Pillars of AI Applications

Moreover, a key focus of Baidu World 2024 was the emphasis on AI agents, which Li described as “the carriers for content, information, and services in the AI-native era.” He believes that AI agents will play a central role in the future of business operations, serving as digital sales representatives, customer service assistants, and more. 

Baidu’s WenXin Agent platform has already attracted the interest of 150,000 businesses and 800,000 developers, showcasing its market potential. Li highlighted various real-world applications of AI agents:

• Qiaocang, an enterprise-facing AI agent, allows businesses to create custom solutions that enhance customer interaction and conversion rates.
• Faxingbao, an industry-specific agent for the legal field, has answered over 16.6 million legal questions, drafted documents, and matched users with lawyers.
• Free Canvas, a collaborative project with Baidu Wenku and Baidu Drive, facilitates the creation of multimodal content by letting users drag and drop various media types onto a digital canvas.

These examples underline Baidu’s commitment to making AI tools integral to multiple industries, from law to project management.

Xiaodu AI Glasses

Adding a tangible element to its AI push at Baidu World 2024, Baidu unveiled Xiaodu AI Glasses, equipped with built-in AI capabilities powered by ERNIE. These smart glasses aim to serve as a “personal assistant,” capable of performing tasks such as snapping photos, summarising content, and translating text in real time. With audio, visual, and location-based support, the glasses can function as a tour guide, provide tailored information, and even curate music playlists to fit the environment.

This launch positions Baidu alongside global tech companies like Meta, which has similarly ventured into AI-assisted eyewear with its Ray-Ban collaboration. However, Baidu’s offering is specifically tailored to the Chinese market, where Meta’s products do not have a foothold due to regulatory restrictions.

ERNIE’s Rapid Growth and Expanding Developer Community

Lastly, a focal point of Baidu World 2024 was the rapid growth of ERNIE’s ecosystem. The foundation model now handles over 1.5 billion API calls daily, up from 200 million per day in May 2023. This spike reflects the increasing adoption of generative AI across various applications in China, where Baidu competes with industry heavyweights like Alibaba and Tencent, as well as rising startups like Moonshot AI and Zhipu AI.

Chief Technology Officer Haifeng Wang reported that ERNIE Bot has accumulated 430 million users, and Baidu’s PaddlePaddle deep learning platform, combined with ERNIE, now has over 18 million developers. These metrics underscore Baidu’s growing influence in China’s AI landscape despite fierce competition from tech giants and AI startups.

Closing Notes

Considering everything, the Baidu World 2024 conference is a powerful statement of Baidu’s ambitions in the AI landscape. The theme, “applications are here,” reflects Baidu’s shift from developing AI technology to deploying it across practical, real-world scenarios.

By addressing longstanding issues like hallucinations in image generation with iRAG and making software development more accessible with MiaoDa, Baidu is positioning itself as a leader in the commercialization of AI. These advancements benefit large-scale enterprises and democratise AI, enabling individuals and smaller businesses to harness its power.

For more tech and AI-related articles, follow our Twitter page for timely updates on the latest announcements across the globe.