How Baidu Uses Advanced AI to Navigate Market Challenges

As the evolution of LLMs progresses, competition intensifies between major technology companies worldwide to develop more sophisticated and capable AI systems.

Over the past year, these models have evolved from primarily language-focused systems to multimodal platforms capable of understanding and generating text, code, images and even video.

Yet Baidu seems to be prepared to push these boundaries further.

In a significant development on the LLM scene, Baidu has announced plans to release the next iteration of its ERNIE (Enhanced Representation through kNowledge IntEgration) foundation model in early 2025.

ERNIE, which has already demonstrated remarkable capabilities in natural language processing and understanding, represents China's most prominent answer to OpenAI's GPT series and Google's PaLM models.

Baidu's continued investment in ERNIE reflects China's broader strategic initiative to develop indigenous AI capabilities, especially in the wake of increased global competition and technology export controls.

This development is also particularly noteworthy given the current geopolitical context, where nations are racing to establish technological sovereignty in AI.

For enterprise customers and developers worldwide, this announcement carries particular weight given ERNIE's growing adoption rates and its demonstrated ability to handle complex tasks across various domains.

The model's success in managing billions of daily API calls has already established it as a crucial tool for businesses looking to integrate advanced AI capabilities into their operations.

ERNIE’s role in Baidu’s development

Despite the challenges reflected in Baidu’s Q3 performance, including “ongoing weakness” in its online marketing business according to Robin Li, Baidu Co-founder and CEO, Baidu's AI capabilities are gaining traction.

"Our relentless focus on cost control helped support our third quarter results. Baidu Core's operating margins remained stable as we managed to further optimise costs and realise more cost efficiencies between business units," says Junjie He, Interim CFO of Baidu.

The adoption of ERNIE has surged significantly.

In November, ERNIE handled approximately 1.5 billion application programming interface (API) calls daily, a dramatic increase from 600 million in August.

This surge indicates growing interest from enterprises and developers in leveraging Baidu's AI technologies for various applications.

Robin points out: “Our strong AI capabilities are gaining broader market recognition, as evidenced by increasing adoption of ERNIE.”

Advancements in AI cloud and ERNIE

Baidu's focus on AI is evident through its commitment to enhancing ERNIE's capabilities.

The latest version, ERNIE 4.0 Turbo, has achieved a 48% improvement in inference throughput compared to its initial release in June.

This improvement allows for faster processing of data and more efficient responses to user queries.

Robin highlights the importance of these advancements: "We expect to launch a new version of ERNIE early next year to further strengthen our leadership position in foundation models."

As Gen AI becomes increasingly relevant across industries, Baidu further aims to cater to diverse user needs while improving performance and accuracy.

The company's AI Cloud division also reported an 11% year-over-year revenue increase, driven by substantial growth in generative AI-related income.

This segment is crucial for Baidu as it seeks to offset declines in online marketing revenue caused by economic pressures affecting advertising demand.

"AI Cloud continued to show healthy and sustainable development in the third quarter”, Junjie summarised. 

Autonomous driving initiatives

In addition to its AI advancements, Baidu is making significant strides in autonomous driving through its Apollo Go service.

RT6 robotaxi
The sixth generation of its autonomous vehicle, known as the RT6 robotaxi, is now operational on public roads across multiple cities in China.

It has fully driverless operations accounting for 70% of rides in Q3, increasing to 80% in October.

Apollo Go
In the third quarter alone, Apollo Go provided 988,000 rides nationwide, marking a 20% increase from the previous year, with cumulative public rides reaching 8 million by October.

Robin noted that fully driverless operations accounted for over 70% of total rides during this period, with that figure increasing to 80% by October.

This progress solidifies Baidu's position as a leader in the global autonomous ride-hailing market. 

“In intelligent driving, Apollo Go continues to lead the global autonomous ride-hailing market, hitting another milestone with our sixth-generation autonomous vehicle now operating on public roads across multiple cities in China”, Robin emphasises.

Baidu's commitment to innovation extends beyond immediate financial results; it reflects a long-term strategy focused on integrating AI into various aspects of its operations.

As Robin summarised during the earnings call: "Despite the near-term pressures, we remain steadfast in our AI-focused strategy and are confident in our long-term trajectory.

“As we further scale AI, we are emboldened to find how it can drive innovations and create value for consumers, enterprises and society at large."

Explore the latest edition of Technology Magazine and be part of the conversation at our global conference series, Tech & AI LIVE.

Discover all our upcoming events and secure your tickets today.

Technology Magazine is a BizClik brand