The Development Trends of Artificial Intelligence in Smart Manufacturing

The Development Trends of Artificial Intelligence in Smart Manufacturing

Smart manufacturing represents the deep integration of advanced manufacturing technology and information technology. It serves as a common enabling technology for the transformation and upgrading of traditional industries and the development of strategic emerging industries in China. Artificial intelligence (AI) is a crucial component of smart manufacturing. How will AI develop in smart manufacturing?

What is Artificial Intelligence?

Artificial intelligence is machine intelligence—systems or disciplines that simulate or mimic human intelligence. The primary research areas of AI include cognitive modeling, knowledge representation, reasoning and application, machine perception, machine thinking, machine learning, machine behavior, and intelligent systems, among others. Reasoning, knowledge, planning, learning, communication, perception, mobility, and manipulation are the foundational topics that AI explores.

Challenges and Trends in AI

Many enterprises aspire to implement AI in actual production applications, and many venture capital funds are keen to invest in AI startups. However, these companies prefer to apply deep learning capabilities directly to solve real-world problems rather than focus on improving deep learning tools, which are often complex and challenging to understand, known only to a few data scientists. This complexity could impede the practical application of AI, emphasizing the need to simplify and productize deep learning platforms.

The State of Artificial Intelligence

Leading providers of deep learning tools, such as Amazon, Microsoft, Google, Intel, and NVIDIA, are attempting to simplify deep learning tools. However, finding a skilled data scientist to streamline these platforms is challenging. Additionally, using these platforms is often complex and costly. Training a single model may take weeks without access to expensive GPU nodes, and many models may not be trainable due to limited hyperparameter optimization.

Current commercialized AI driven by deep learning is mostly confined to two main areas: text and speech processing, and image and video processing. These fields are commercially viable, with AI and deep learning being actively adopted.

Artificial Intelligence Applications in Manufacturing

AI will continue to manifest primarily in the form of natural language processing chatbots. In 2015, only 25% of companies had heard of chatbots, but by 2017, a significant portion was planning to build them. Smart speakers like Amazon Echo represent AI's entry into the consumer domain, with sales exceeding 20 million. The third quarter of 2017 saw a 708% surge in smart speaker shipments, signaling a domestic market boom. Chinese companies like Alibaba and Xiaomi have launched their smart speaker devices, competing with low-cost market strategies.

Voice recognition is expected to continue its rapid growth, potentially becoming a standard user interface across all systems. As visual entry points become saturated, audio devices will play an increasingly significant role, and voice-based AI will find its way into more smart products like speakers, TVs, refrigerators, cars, and wearable devices.

AI Driving Technologies

The actual application of deep learning AI in image and video recognition is relatively limited, with facial recognition being the only significant market prospect. Although autonomous driving is a hot area for AI startup investment, traditional automakers and tech companies continue to invest, there are questions about whether autonomous vehicles are primarily aimed at end-consumers or transportation and logistics companies. The market potential requires further validation and exploration.

The Dual-Edged Nature of AI

In 2017, China's information industry turned to AI, with news apps like Toutiao using online tracking and location tracking algorithms to deliver personalized content. This approach allows targeted advertising, reducing advertising costs and minimizing ineffective advertising for readers.

However, AI poses privacy risks. Although users can opt out of tracking, apps often know users better than friends or family. With the proliferation of facial recognition cameras, our faces may be recorded and stored in facial recognition databases. If this data falls into the wrong hands, the implications could be dire.

Smart Manufacturing's Path Forward

Smart manufacturing is based on human-created intelligence rather than solely AI. Industrial intelligence derived from years of accumulation in the industrial domain and AI from the information domain must integrate and learn from each other. These two main streams, along with other smart automation technologies, represent the mainstream trend in smart manufacturing. Starting from industrial intelligence and gradually incorporating AI might be the path for Chinese enterprises in their smart manufacturing journey.