Baidu bolsters AI lineup with text-to-image generator, no-code app builder

By Liam Mo and Brenda Goh

BEIJING (Reuters) – China’s Baidu Inc (NASDAQ:) unveiled a slew of new applications for its artificial intelligence technology on Tuesday, including a text-to-image generator and a tool that enables users to develop software applications without coding expertise.

The country’s leading search engine company is among tech firms shifting their focus to the commercialization of large language model (LLM) applications after nearly two years of heavy investment in research and development in models that they tout as alternatives to OpenAI’s GPT.

At the annual Baidu World Conference, CEO Robin Li introduced I-RAG, a text-to-image technology that leverages Baidu’s search capabilities to address the “hallucination” issue, referring to the generation of images that deviate from the input text or contain non-existent elements.

Li said the company’s Ernie platform now handles 1.5 billion user queries and interactions daily, a significant increase from the 200 million daily requests reported in May.

These interactions include tasks such as generating text, answering questions, and assisting with various applications utilizing AI.

Baidu’s commercialization efforts have largely revolved around its AI agents, which allow users to create their own applications.

The company has also integrated the technology into its existing product lineup and offering it to external users via its cloud services.

Baidu also unveiled a pair of glasses featuring a built-in AI assistant, developed by it hardware division, Xiaodu. The gadget is equipped with cameras to capture photos and videos and supports voice interactions powered by Ernie.

The company is not aiming to create a “super app” based on AI, Li said, signalling a divergence in strategy from other companies such as ByteDance, which has launched multiple standalone AI apps this year.

Baidu also introduced Miaoda, which utilizes its LLM capabilities to generate code, allowing users to develop software applications without extensive coding expertise.