[Answer] How Should We Evaluate Xiaomi’s AI Smart Speaker, “Xiao Ai”?

[Answer] How Should We Evaluate Xiaomi’s AI Smart Speaker, “Xiao Ai”?

The rapid growth in smart speaker sales is a result of progress in artificial intelligence technology, and to a significant extent, it is also one of the major signs of AI’s broader breakout. Compared with traditional speech recognition technology, models trained through large-scale deep learning have made enormous advances in voice recognition. Smart speakers, meanwhile, rely on cloud computing to handle the core recognition and functionality in the cloud (I am not a smart speaker developer myself, so this is just my inference). This has led to a substantial drop in speaker costs compared with the past, while performance has improved dramatically. A standalone computer or embedded system would basically have no way of reaching the current level. This is the main reason the smart speaker industry has developed so rapidly.

Another factor is the rapid development of the Internet of Things. As IoT technology has advanced quickly, the cost of smart home products has fallen significantly. This has made it very easy to expand the functionality of smart speakers. People can now conveniently upgrade household appliances through various methods to give them smart capabilities, without requiring large-scale investment or replacement.

At present, the main limitations facing the smart home industry, represented by smart speakers, still lie in network stability and bandwidth. However, with the rollout and wider adoption of 5G technology, IoT is bound to develop even faster. At the same time, as smart speakers continue to spread, the amount of data and the number of training samples will increase rapidly, which will inevitably lead to qualitative improvements in model training. Therefore, the explosive growth of smart speakers is only the first step. From the perspective of the smart speaker industry, this is the first step in smart speakers and smart homes entering and taking over the household. From the perspective of the AI industry, Andrew Ng has predicted that from now until 2030, artificial intelligence could contribute about 13 trillion US dollars in economic output. So this is also the beginning of AI entering a phase of large-scale acceleration.