Key Points
- Baidu’s Text-to-Image generator I-RAG tool uses search capabilities to improve accuracy in AI-generated images, tackling the “hallucination” issue.
- The Ernie platform now handles 1.5 billion daily interactions, supporting various AI applications, from text generation to answering questions.
- Baidu’s Xiaodu division introduced smart glasses with an AI assistant integrating photo, video, and voice capabilities.
- The no-code development tool Miaoda enables users to develop applications without coding knowledge, expanding software creation to a broader audience.
China’s Baidu Inc. introduced a suite of new artificial intelligence (AI) applications at its annual Baidu World Conference, with innovations designed to boost accessibility and usability across various sectors. Among the key announcements were a text-to-image generator called I-RAG and a Miaoda tool, which enables users to create software applications without coding expertise. The new offerings are part of Baidu’s pivot toward commercializing large language model (LLM) applications, building on years of investment in AI research to provide alternatives to tools like OpenAI’s GPT.
Baidu’s CEO, Robin Li, introduced I-RAG, a cutting-edge text-to-image tool that Baidu claims addresses a common issue in AI-generated images known as “hallucination,” where visuals generated deviate from the input text or contain elements that don’t exist. By integrating Baidu’s robust search capabilities into this technology, the company aims to ensure greater accuracy and relevance in AI-generated images.
Li reported that Baidu’s AI-driven Ernie platform now processes approximately 1.5 billion user queries and interactions daily, substantially increasing from the 200 million daily requests recorded in May. The Ernie platform supports various AI applications, including generating text, answering questions, and powering AI-driven tools individuals and businesses use.
Baidu’s recent efforts focus on creating AI agents that allow users to develop their applications. This aligns with Baidu’s shift from pure AI research to practical applications in commercial settings. The company has already integrated these AI tools across its existing product offerings and made them available to external clients via its cloud services.
As part of its hardware advancements, Baidu also unveiled a pair of smart glasses developed by its hardware division, Xiaodu. These glasses feature a built-in AI assistant and integrated cameras for capturing photos and videos. They also support voice interaction powered by the Ernie platform. This new gadget aims to make AI a more natural part of everyday life by providing real-time assistance on the go.
Baidu intends to offer modular tools that users can incorporate into their existing systems, reflecting a flexible, user-focused approach. One of the most significant developments from the conference was Miaoda, a tool designed to democratize software development. Utilizing Baidu’s LLM technology, Miaoda allows users to generate code without advanced programming skills, thus broadening access to software creation.
