Artificial intelligence for the first time exceeds human eye accuracy. Face recognition accuracy has increased by 4 orders of magnitude.

During his speech titled “The Present and Future of AI Empowerment” at the MIT Technology Review Summit, Wang explained how AI is transforming everyday life through cloud computing and edge devices. He emphasized that AI is no longer just a futuristic concept but a powerful tool shaping our future.

Over the past decade, deep learning has been the key driver behind the rapid development of AI. According to Wang, three main factors have fueled this progress: the availability of big data, increased computational power through cloud services and specialized hardware like GPUs and AI chips, and advancements in deep learning algorithms.

He gave an example: in 2014, machine face recognition had an error rate of one in ten thousand, but now it’s as low as one in a billion. This means that while earlier systems were like a 4-bit password, today’s systems are more secure, equivalent to an 8-bit password. This dramatic improvement shows how far the technology has come in a short period.

As algorithms continue to evolve, their applications are expanding. Initially used for 1:1 identity verification, AI now supports dynamic monitoring, such as identifying suspects in real-time. Today, it's possible to search through hundreds of millions of images across a city and reconstruct human movements with high precision. This level of accuracy is largely due to the power of deep neural networks, which can be trained on thousands of layers to outperform human vision.

Neural networks have grown significantly, from just five layers in 2012 to over 1,200 layers today. However, despite this complexity, front-end applications often use smaller networks. Why is that? Wang explained that to create a small but efficient network, a much deeper and stronger model must first be trained. He likened this process to a primary school student learning from a teacher who has extensive knowledge.

In this analogy, the teacher represents the deep neural network—capable of absorbing vast amounts of information from large datasets. Once the teacher has learned everything, it can pass its knowledge down to a smaller, more practical network, enabling it to perform well in real-world scenarios.

Improvements in AI algorithms have led to numerous real-world applications, enhancing public safety. For instance, criminals who had been anonymous for over a decade were eventually caught using dynamic face recognition systems, even after changing their ID cards. Additionally, the police have successfully located missing elderly people and children by combining city-wide camera networks with facial recognition technology.

According to Shangtang’s research, in 2014, training a system with 200,000 faces achieved a 98.5% accuracy rate, while humans achieved 97.5%. By 2015, with 300,000 faces, accuracy rose to 99.55%. In 2016, training on 60 million faces reduced the error rate to one in a million, and by 2017, with 2 billion faces, the error rate dropped to one in 100 million. These results show how AI is now being applied across various industries with high reliability.

On the hardware side, Wang also mentioned the strategic partnership between Shangtang and Qualcomm. He believes that for AI to become mainstream, it must be integrated into front-end devices. These devices rely heavily on specialized chips. Currently, Shangtang’s face unlock technology is used by over 100 million mobile phone users, all of which depend on Qualcomm’s chip support.

Coaxial Cable

Coaxial Cable RG59 RG6

Conductor: Bare copper or Copper Clad Steel (Solid)

Insulation: Foamed PE

Shield: Aluminum/PET-Foil Bonded

Sheath: PVC (Polyvinyl Chloride)

Applications: Those Coaxial Cables are suitable for Digital, Analog, CATV, Satellite, Off-Air, Broadband

Coaxial Cable, Digital Coaxial Cable, RG59, RG6, CCTV Cable

Shenzhen Bendakang Cables Holding Co., Ltd , https://www.bdkcables.com