While the US, with its line-up of powerful tech companies, such as OpenAI, Microsoft, and Google, seems to be spearheading the AI race, a force from the Far East is quietly establishing its dominance. China’s multinational giant, Baidu, has been making strides in AI that others should probably worry about.
A few days ago, at the Baidu World 2024 conference in Shanghai, the big-tech company unveiled a number of AI products that catered to enterprises and consumers alike. AI Agents, image generation, smartglasses, name anything and Baidu has it covered.
Taking on Hallucinations
When the West is flaunting its text-to-image models such as DALL·E 3, Gemini, and others that have landed themselves in trouble for spewing misinformation, Baidu has pretty much struck the right chord.
Addressing the hallucinations that tend to mar the quality of AI-generated output, Robin Li, the founder and CEO of Baidu Inc, unveiled iRAG, an image-based retrieval augmented generation tool to tackle hallucinations in image generation.
Li highlighted that by utilising Baidu Search’s extensive database of hundreds of millions of images and the company’s advanced foundation model capabilities, iRAG enables text-to-image models to produce hyper-realistic visuals while lowering image-production costs.
Baidu’s iRAG output. Source: Baidu World 2024
On the application part, Baidu looks to enhance the usability of text-to-image outputs across various visual mediums, such as comics, storyboards, and posters. According to Li, minimising hallucinations is setting the stage for rapid growth in AI applications.
Developer-Focussed Too
To tackle hallucinations is one thing, but to enable developers is huge. Baidu’s Miaoda, a no-code tool, enables users to build applications simply by describing them in natural language without needing to write a single line of code.
No-code platforms have been a constant offering by all major tech companies. Google’s AppSheet, Microsoft’s Power Apps, and Amazon’s Honeycode are just some of the big names offering such capabilities. Enterprise biggies, including Oracle and Salesforce, are also offering the same with agentic capabilities.
At the recent Oracle CloudWorld and Dreamforce 2024 events, both companies unveiled 50-100 AI agents across their suite of enterprise products. But just when you think this has been an emerging trend with the West, Baidu proves you wrong.
Its no-code tool, Miaoda, incorporates multi-agent collaboration, leveraging ERNIE’s capabilities for task coordination and multi-tool invocation, which taps into Baidu’s expansive suite of APIs, including web search, mapping, and image generation.
The tool’s ability to call upon different agents from domains such as project management, content editing, programming, and quality assurance allows users to streamline workflows and execute complex processes without technical expertise.
So far, Baidu’s ERNIE AgentBuilder has attracted the attention of 150,000 businesses and 800,000 developers. The top 100 agents include character-based agents such as the ‘Farmer Academician’, along with agents tailored for tools, industries, workplaces, emotions, and entertainment.
“Today, while all leading global tech firms are paying attention to agents, few have made them as central to their strategy as Baidu has,” said Li. “Agents are more human-like, more intelligent, and act like your sales and customer service representatives, or assistants. Agents will become a new vehicle for content, information and services.”
There’s AI Glasses Too
Just when you thought Meta’s AI smartglasses could replace smartphones, Baidu comes up with its version of the AI wearable.
Ying Li, the CEO of Xiaodu Technology (a product line of Baidu), took the stage to introduce Xiaodu AI Glasses, a new device that combines visual, audio, and location-based capabilities to function as a versatile AI assistant.
Powered by ERNIE and equipped with advanced language and image processing capabilities, Xiaodu AI Glasses can act as a virtual tour guide, offering real-time location-based information through Baidu Maps, provide instant translations, and even assist with intelligent note-taking and content summarisation.
These new AI glasses are likely to be made available in the first half of 2025.
China Catches Up
A few months ago, Taiwanese computer scientist, entrepreneur, and author of ‘AI Superpower’, Kai-Fu Lee, said that China will catch up with the US in AI capabilities and is just 6-9 months behind the US.
Let’s take autonomous vehicles, for instance. Baidu has been going strong with Apollo, its autonomous project, covering major cities in China at L4 autonomy. It is going against some incumbent players like Waymo and Tesla Robotaxi. Furthermore, the company has been a key platform provider for a number of automotive companies.
On the computing side, Baidu recently introduced Baige 4.0, a new version of its AI Heterogeneous Computing Platform, which focuses on enhancing cluster stability and efficiency.
It’s interesting to note that while the US ponders over sanctions, with the recent development of TSMC being banned from sending AI chips to China and a potential trade war with the new Trump government, it is obvious that none of this will hamper China’s AI developments. Reports even claim that GPU-giant NVIDIA is preparing a specialised AI chip for the Chinese market.
The post The Silent AI Leader Nobody is Talking About appeared first on Analytics India Magazine.
