A visitor tries out a smart speaker at a high-tech expo in Jinhua, Zhejiang province.
Xiaodu, hearing her commands, switches itself on and begins to play Zhang's favorite music.
In Beijing, when she asks what the weather will be like the next day, the voice-activated digital assistant says the temperature will be 21C in the morning and that she needs to take a coat.
Today, customers like Zhang are able to watch videos, view photos, make video calls and search the internet on their smart speakers using their voice or hands. Then, they can order meals, groceries, and medicine with voice commands.
Smart speakers are becoming more popular in China due to the rapid advancement of voice recognition and language comprehension technologies, resulting in a surge in popularity among young tech-savvy consumers.
Experts said the integration of fast-growing generative AI technology and chatbot abilities into voice assistants will significantly enhance smart speaker functionality, positioning them as valuable companions within homes.
Baidu Inc, a tech giant in China, is focusing on the growth of the smart speaker market in China and is investing in research to develop intelligent speech interaction technologies.
The company has unveiled a series of intelligent video speakers, Xiaodu Zaijia, equipped with built-in touch screens and powered by DuerOS, Baidu's conversational AI technology similar to Amazon's digital assistant Alexa.
Jing Kun, vice-president of Baidu and CEO of Xiaodu Technology -Baidu's smart speaker unit - said the penetration rate, development speed and the number of smart speakers with built-in touch screens will surpass that of traditional smart speakers.
Jing said the company wants to allow smart assistants to be involved in different scenarios and play a greater role in people's lives. By using AI conversation technology, smart display speakers can connect with seniors with immediate assistance and serve as always-on virtual companions, enhancing their quality of life.
In the future, Jing said Baidu's AI-large language models, or LLMs, will be applied to the company's intelligent terminal devices, such as smart speakers, tablets and headsets.
LLMs are computer algorithms that possess massive amounts of data and can generate content such as images, text, audio, and video. The technology underpinning ChatGPT, an AI chatbot developed by OpenAI, is crucial to its success.
Data from consultancy Runto showed that the sales of smart speakers in China will go up from 26.31 million units last year to 27.15 million units in 2023, a slight increase of 3 percent year-on-year.
Xiaodu took the lead and dominated the domestic smart speaker market in 2022, holding a 35 percent share of the market. Xiaomi ranked second, 31 percent, followed by Alibaba's Tmall Genie and Huawei.
Advancements in speech recognition, NLPs and LLMs will greatly improve voice assistants' functionality and user experience, and generative AI technology can provide voice assistants with a far superior cognitive comprehension of user intentions, according to a report published by Canalys.
These advancements have the potential to transform voice assistants into indispensable virtual companions that provide convenience and companionship, surpassing the currently limited capabilities and overcoming frustrations voiced by users, it said.
In addition to its award-winning smart speaker, Tmall Genie, Alibaba Group has made significant investments in improving its AI-of-things ecosystem. Tmall Genie provides a broader set of content and services across Alibaba's platforms, encompassing entertainment, education, healthcare, and online shopping.
It has heightened efforts to improve research into IoT operating systems, edge computing, and AI technologies like natural language processing. These technologies will help elevated the Tmall Genie user experience by enabling more personalized, natural voice and visual interactions, Alibaba said.
In addition, Alibaba announced it plans to add its ChatGPT-like AI chatbot into all of its applicable businesses, including Tmall Genie and DingTalk, the company's workplace messaging app.
Low said the smart assistant ecosystem needs to be supported by third-party developers, hardware vendors and service providers, adding that ecosystem partners must pay close attention to what platform vendors achieve while having a keen sense for uncovering new uses as users' habits change.