A desert robot depicts AI’s vast opportunities

When Hongzhi Gao was young, he lived with his family in Gansu, a province located in the center of northern China through the Tengar Desert. Thinking back to his childhood, he remembers the constant, constant dust storm outside his house, and it didn’t take him more than a minute to get out most months of the year, before the sand filled any empty space and slipped inside his house. . Pockets, shoes and his mouth. The monotony of the desert stuck in his head for years, and at university he turned that memory into the idea of ​​building a machine that could bring plant life into the desert landscape.

Efforts to prevent desertification પ્રક્રિયા the process by which fertile soil becomes desert-have focused primarily on expensive manual solutions. Hongzi designed a robot with deep learning technology to automate the planting process: from identifying the best places to planting tree seedlings. Despite having no experience with AI, as an undergraduate student, Hongzi used Baidu’s Deep Learning Platform PaddlePaddle to combine different modules to create a robot with better object detection capabilities than similar machines already available on the market. It took Hongzi and his friends less than a year to spin the final product and get it working.

Hongzi’s desert robot serves as an example of the increasing accessibility of artificial intelligence.

Today, more than four million developers are using Baidu’s open source AI technology to create solutions that can improve people’s lives in their communities, and many of them have little or no technical expertise in the field. “Over the next decade, AI will be the source of the changes taking place in every fabric of our society, which will change the way industries and businesses operate. Technology will take us deeper into the digital world and expand the human experience, “said Robin Lee, CEO of Baidu at the Baidu Create 2021, AI Developer Conference.

As we enter a new chapter in the evolution of AI, Baidu’s CTO, Haifeng Wang, identifies two key trends that move the industry forward: AI will continue to mature and increase its technological complexity. And at the same time, the cost of deployment and barriers to entry will decrease – benefiting both the enterprise creating AI-powered solutions on a scale and software developers exploring the world of AI.

Merging of knowledge and data with deep learning

The integration of knowledge and data with Deep Learning has significantly improved the efficiency and accuracy of AI models. Since 2011, Baidu’s AI infrastructure has been acquiring and integrating new information into a large-scale knowledge graph. Currently, this knowledge graph contains over 550 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.

This knowledge graph and huge data points together form the building blocks of Baidu’s newly released pre-trend language model PCL-BAIDU Wenxin (version ERINIE 3.0 Titan). This model outperforms other language models without knowledge graphs on 60 Natural Language Processing (NLP) functions, including reading comprehension, text classification, and semantic similarity.

Learning throughout the methods

Cross-model learning is a new field of AI research that seeks to improve the cognitive understanding of machines and better mimic the adaptive behavior of humans. Examples of research efforts in this area include automated text-to-image synthesis, where a model is trained to create images from a single text description, as well as algorithms designed to understand visual content and express that understanding with words. The challenge with these tasks is for machines to make meaningful connections across different types of datasets (e.g., images, text) and to understand the interdependence between them.

The next step for AI is to merge AI technologies such as computer vision, speech recognition and natural language processing to create a multi-modal system.

On this front, Baidu has introduced a variety of its NLP models that combine language and visual semantic understanding. Examples of real-world applications for this type of model include digital avatars that can understand their surroundings like humans and handle customer support for businesses and algorithms that can “draw” pieces of art and create poems based on their understanding of the generated artwork. Can write .

There are even more creative, impressive potential outcomes for this technology. The pedalpad platform can make meaningful connections between vision and language, which is why a group of master’s students in China created a dictionary to save endangered languages ​​in regions such as Yunnan and Guangxi by translating them more easily into simple Chinese.

AI integration across software and hardware and industry-specific use cases

As AI systems are increasingly applied to solving complex and industry-specific problems, the emphasis is on optimizing software (deep learning framework) and hardware (AI chip), rather than individually optimizing each one. Considering such factors. As computing power, power consumption and latency.

Next, there is tremendous innovation at the platform level of Baidu’s AI infrastructure, where third-party developers are using deep learning capabilities to create new applications tailored to specific use cases. The PaddlePaddle platform has a range of APIs to support AI applications in new technologies such as quantum computing, life sciences, computational fluid mechanics and molecular dynamics.

There are also practical uses for AI. In Shoguang, a small town in Shandong Province, for example, AI is being used to streamline the fruit and vegetable industry. Dozens of vegetable sheds require only two people and one application to operate.

And it’s worth noting, says Wang, “Despite the increasing complexity of AI technology, open-source deep learning platforms bring together applications such as processors and operating systems, reducing the barriers for companies and individuals seeking to incorporate AI into their business.”

Reduced access barrier for developers and end users

On the technology front, pre-training of large models such as PCL-BAIDU Wenxin (version ERNIE 3.0 Titan) has overcome many of the common barriers faced by traditional models. For example, these general-purpose models have helped lay the groundwork for running a variety of downstream NLP functions, such as text classification and question-answer, in an integrated space, whereas in the past, each type of task had to be solved. By a different model.

PaddlePaddle also has a range of developer-friendly tools, such as modal compression technology to fit general purpose models into more specific use cases. The platform provides an officially supported library of industrial-grade models with over 400 models, from large to small, that maintains only a fraction of the size of general-purpose models but achieves comparable performance by reducing model development and deployment costs. .

Today, Baidu’s open source deep learning technology supports a community of over four million AI developers who have collectively created 476,000 models, contributing to the AI-driven transformation of 157,000 businesses and organizations. The examples described above are the result of innovations happening across all levels of Baidu AI infrastructure, integrating technologies such as voice recognition, computer vision, AR / VR, knowledge graphs and pre-training large models that are one step closer to understanding. A world like humans.

In its current state, AI has reached a level of maturity that allows it to perform wonderful functions. For example, the recent launch of the Metaverse XiRang would not have made it possible for participants around the world to create digital avatars to connect to their devices without a pedalpad platform. In addition, future breakthroughs in areas such as quantum computing could significantly improve the performance of metavaruses. This shows how Baidu’s various offerings are intertwined and interdependent.

In a few years, AI will be closer to the root of our human experience. What was like steam power, electricity and internet for previous generations will be for our society. As AI becomes more complex, developers such as Hongzi will work harder in the capacity of artists and designers, giving creative freedom to explore use cases previously only considered theoretically possible. The sky is the limit.

This content was created by Baidu. It was not written by the editorial staff of MIT Technology Review.

Similar Posts

Leave a Reply

Your email address will not be published.