The center of northern China by the Tengger Desert is where Hongzhi Gao lived with his family when he was young. When he was a kid, he remembers the constant, steady wind of dirt outside his house and how quickly sand would fill his pockets, boots, and mouth. The monotony of the desert stuck in his head for a long time, and at university he came up with an idea to build a machine that could bring plant life to the desert landscape.
Efforts to stop desertification have been focused on expensive manual solutions. Hongzhi designed a robot with deep learning technology to automate the process of tree planting. Despite having no experience with artificial intelligence, an undergraduate student Hongzhi used a deep learning platform to build a robot with better object detection capability than similar machines already available in the market. It took less than a year for Hongzhi and his friends to finish their project.
The increasing accessibility of artificial intelligence can be seen in Hongzhi's desert robot.
More than four million developers are using the open source artificial intelligence technology from Baidu to build solutions that can improve the lives of people in their communities, and many of them have little to no technical expertise in the field. Within the next decade, artificial intelligence will be the source of changes taking place across every fabric of our society. Robin Li, the CEO of Baidu, said at the conference that the technology will expand the human experience by taking us on a deeper dive into the digital world.
As we enter a new chapter in the evolution of artificial intelligence, the CTO of Baidu identified two key trends that underpin the industry's path forward. The cost of deployment and barrier to entry will decrease, which will benefit both enterprises and software developers.
Merging of knowledge and data.
The integration of knowledge and data with deep learning has improved the efficiency and accuracy of models. In the past five years, the artificial intelligence infrastructure of Baidu has been acquiring and integrating new information into a large-scale knowledge graph. This knowledge graph has more than 500 billion facts, covering all aspects of everyday life, as well as industry-specific topics, including manufacturing, pharmaceuticals, law, financial services, technology, and media and entertainment.
The building blocks of the new pre-trained language model PCL-BAIDU Wenxin are made up of a knowledge graph and massive data points. The model beats other models without knowledge graphs on 60NLP tasks, including reading comprehension, text classification, and semantic similarity.
Learnings across different types of therapies.
Cross-modal learning is an area of artificial intelligence research that seeks to improve machines' cognitive understanding and mimic humans' adaptive behavior. Automatic text-to-image synthesis, where a model is trained to generate images from text descriptions alone, is one example of research efforts in this area. The challenge with these tasks is for the machines to understand the interdependencies between different types of data.
The next step in the development of artificial intelligence is to combine technologies like computer vision, speech recognition, and natural language processing to create a multi-modal system.
The variant of the model that is rolled out by Baidu ties together language and visual understanding. Examples of real-world applications for this type of model include digital avatars that can perceive their surroundings like human beings and handle customer support for businesses, as well as an application that can draw pieces of art and compose poems based on their understanding of the generated artworks.
There are more possibilities for this technology. A group of master's students in China created a dictionary to preserve the languages of Yunnan and Guangxi, which are in danger of being lost, because of the PaddlePaddle platform.
There is an integration of artificial intelligence across software and hardware.
The software and hardware of the system should be adjusted as a whole, instead of individually, taking into account factors such as computing power, power consumption, and latency.
Third-party developers are using the deep learning capabilities to build new applications tailored to specific use cases at the platform layer of the artificial intelligence infrastructure. The PaddlePaddle platform has a number of applications that are supported by the platform's DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch DropCatch
There are practical uses for artificial intelligence. The fruit and vegetable industry is being streamlined in a small city in the province. Two people and an app are all it takes to manage vegetable sheds.
Wang says, "Despite the increased complexity of artificial intelligence, open-source deep learning platform brings together the processor and applications like an operating system, reducing barriers to entry for companies and individuals looking to incorporate artificial intelligence into their business."
Developers and end users have a reduced barrier to entry.
Pre-training large models like PCL-BAIDU Wenxin have solved many of the problems faced by traditional models. In the past, each type of task would have to be solved by a separate model, but with these general-purpose models, each type of task can be run in one consolidated place.
There are a number of developer-friendly tools in PaddlePaddle, such as model compression technologies to modify general-purpose models to fit more specific use cases. The platform provides an officially supported library of industrial-grade models with more than 400 models, ranging from large to small, which retain only a fraction of the general-purpose models' size but can achieve comparable performance, reducing model development and deployment costs.
More than four million developers have created 476,000 models using the open source deep learning technology from Baidu. The examples are a result of innovations happening across all layers of the artificial intelligence infrastructure, which integrates technologies such as voice recognition, computer vision, augmented reality, knowledge graphs, and pre-training large models that are one step closer to seeing the world like humans.
The current state of the machine allows it to do amazing tasks. The recent launch of Metaverse XiRang would not have been possible without the help of PaddlePaddle. The performance of metaverses could be improved by quantum computing. This shows how different offerings are interdependent.
In a few years, the core of our human experience will be artificial intelligence. It will be to our society what the internet was to previous generations. Hongzhi will be able to explore use cases previously considered only theoretically possible as the complexity of artificial intelligence increases. The sky is not enough.
The content was produced by a company. It was not written by the editorial staff of MIT Technology Review.
