Key Takeaways:
- Baidu’s founder, Robin Li, has been focused on artificial intelligence (AI) since his undergraduate days, despite being advised against it by his professor.
- Li believes that the global AI market will continue to grow, with the foundation model space eventually coalescing around a few champions.
- Baidu is taking an application-driven approach to developing its AI models, focusing on areas such as search, digital humans, and robotaxis.
- The company is expanding internationally, with its Apollo Go robotaxis being deployed in the Middle East and Europe.
- Li thinks that the value in chips is often misguided, and that the application value-add needs to increase in order to sustain investment in AI.
- Baidu is embracing an open-source model, but Li notes that open source is not the key to value creation.
Introduction to Baidu and AI
The story of Baidu, China’s tech giant, is one of innovation and perseverance. Founded by Robin Li in 2000, the company has grown to become a $50 billion entity, with a strong focus on artificial intelligence (AI). Li’s interest in AI dates back to his undergraduate days at Peking and Tsinghua universities, where he took classes on the subject. However, it wasn’t until 2010 that Baidu started to invest in AI, recognizing its potential to improve search engine results. Today, Baidu is one of China’s top full-stack AI companies, offering a range of products and services from chips and cloud infrastructure to models, agents, and consumer products.
The Growth of AI
Li believes that the global AI market will continue to grow, driven by the increasing adoption of AI in various sectors. He notes that 2025 will be a pivotal year for AI adoption, with the foundation model space expected to coalesce around a few champions. Baidu is well-positioned to take advantage of this trend, with its Ernie Bot 5.0 competing well with other large language models. Li attributes Ernie Bot’s success to its application-driven approach, which focuses on areas such as search, digital humans, and robotaxis. He notes that no foundation model can be better than others in every aspect, but by optimizing for specific applications, Baidu can create more value.
International Expansion
Baidu is expanding internationally, with its Apollo Go robotaxis being deployed in the Middle East and Europe. Li notes that the company is taking a flexible approach to expansion, partnering with local companies and regulators to ensure that its technology is deployed safely and efficiently. He also highlights the advantages of China’s supply chain, which enables Baidu to manufacture robotaxi cars at a lower cost than Western companies. Despite the challenges of regulatory and geopolitical issues, Li is confident that Baidu’s technology is ready for mass deployment.
Chips and Applications
Li thinks that the value in chips is often misguided, and that the application value-add needs to increase in order to sustain investment in AI. He notes that the current pyramid structure, where most value is realized at the chip level, is unhealthy. Baidu is taking a different approach, focusing on applications and creating value at the application layer. Li believes that this approach will enable the company to create more value and drive growth in the AI market.
Open-Source Model
Baidu is embracing an open-source model, with its PaddlePaddle deep learning framework rivalling TensorFlow and PyTorch. Li notes that open source is not the key to value creation, but it can help to drive innovation and adoption. He also highlights the importance of creating value at the application layer, rather than just focusing on models and chips.
Regulatory Challenges
Li notes that regulatory challenges are a major hurdle for AI adoption, particularly in China. He believes that the Chinese government is pro-innovation, but also needs to balance the needs of different stakeholders. Li thinks that the U.S. approach to regulation, which focuses on safeguards and guardrails, is not necessarily the right approach. Instead, he advocates for a more nuanced approach that balances innovation with regulation.
Job Displacement and Energy Consumption
Li acknowledges that AI may displace some jobs, but believes that it will also create new opportunities. He notes that Baidu has helped to create data labelling centers that hire thousands of people, and that the company is working to create new job opportunities in areas such as AI training and development. Li also highlights the importance of addressing energy consumption, which is a major challenge for AI adoption. He notes that Baidu has made significant efforts to reduce its energy consumption, and that the company is working to develop more energy-efficient models and technologies.
The Future of AI
Li is cautious about predicting the future of AI, noting that the technology is evolving rapidly. He believes that AI will continue to transform society in unexpected ways, and that it is difficult to imagine what the future will hold. However, he is excited about the potential of AI to drive innovation and growth, and is committed to ensuring that Baidu remains at the forefront of the AI revolution.