Tianjic, a Hybrid Chip Design That Could Be the Future of Thinking Machines

Tianjic, a Hybrid Chip Design That Could Be the Future of Thinking Machines

Amid the US-China tech and trade war underway, China, the second largest economy of the world is fully geared up towards advancements in the areas of artificial intelligence and semiconductor design.

In accordance to the plan of action issued by the State Council in July 2017, the country is set to achieve global dominance in AI by 2030 and to grow its domestic industry to about 1 trillion yuan.

In 2018, it was an alarming situation when China’s annual chip imports surged to US$312 billion, exceeding the value of its oil imports.

China falls 10-years behind in manufacturing of integrated chips despite being on the same level with its international counterparts in chip designing. To eliminate dependency on foreign chip technology, an investment of 200 billion yuan was made towards in-house semiconductor development.

Up until now, the research conducted in the fields of AI had been limited to domain-dependent and problem-specific solutions. However, the focus has been shifted to AGI in an attempt of creating general-purpose systems with the intellect similar to the human mind.

Professor Shi Luping of Tsingha University led a group of researchers to develop the Tianjic chip, which brought the world one step closer to achieving artificial intelligence (AGI), also known as “full AI”.

The chip was made by integrating the “computer-science-oriented and neuroscience-oriented” approaches to developing AGI. This enabled the chip to run diverse machine-learning algorithms and reconfigurable building blocks, among others through a single computing platform.

The chip was tested with the help of a self-driving bicycle to assess its performance in a road test. It proved to be successful in supporting multiple coding schemes and adaptive capabilities in complicated situations as it was able to detect and avoid obstacles, maintain balance, recognize voice commands, track and make decisions in different scenarios.

According to Shi Luping, a version of the chip is planned to be launched in the early next year for mass production. Potentially the chip for general AI could be utilized across several industries. The chip could be of use in the fields of autonomous driving, robotics and automation as well.

The measures by incubate start-ups to focus on the development of applications based on the Chinese researchers’ work on the Tianjic chip design are now in progress.