China’s JD embarks on its ChatGPT variant amidst local regulatory concerns

will be putting more focus on serving businesses, rather than consumers. Interestingly, Nikkei recently reported that regulators in China told Chinese companies not to offer ChatGPT services to the public amid growing alarm in Beijing over the AI-powered chatbots’s uncensored replies to user queries. JD.com declined to comment when asked about this development. The application “The industrial ChatJD application will build a human-computer dialogue platform that understands and generates tasks in natural language processing, with parameters estimated to total hundreds of billions,” a JD spokesperson told Inside Retail. The ‘Yanxi AI’ platform, which was developed from JD’s early smart customer service chatbot, applies dialogue generation in both customer service and generating marketing content like selling points.  Yanxi is currently used by 588 million consumers and 178,000 merchants on JD. It has over 40 independent systems, 3,000 intentions and 30 million high-quality Q&As that cover over 10 million retail commodities. For example, the chatbot service can predict a customer’s purchase intention, provide pre-sale consultation, recommend products based on user scenarios and more. It is essentially equipped with the ability to guide shoppers. The big picture One example of the app’s prowess is in the field of text generation. It can auto-generate short- and long-form copy for businesses to use in marketing.  According to the spokesperson, it has covered more than 3,000 product categories, cumulatively generated 3 billion characters, and has brought in more than RMB 300 million yuan in sales. It has also played a role as an AI-powered ambassador in widely used e-commerce live streaming for more than 4,000 brands. “Our intelligent chatbot has provided services for hundreds of thousands of merchants, saving more than 30 per cent in labour costs for merchants. The service has covered over 80 per cent of categories in the retail industry,” the spokesperson elaborated. Brands like Midea, Adidas and Lenovo have all had the treatment, and it seems that every day on JD.com, 10 million intelligent interactions are generated, accumulating in a knowledge graph covering vertical fields such as e-commerce, finance, insurance, logistics. Experts weigh in “In coming months, the industry will continue to witness a number of platforms integrating their own versions of ChatGPT, including chatbots focused on verticals such as retail, finance and others,” Charlotte Dunlap, research director at GlobalData, told Inside Retail. She said that as the most advanced AI-powered chatbot technology the world has ever seen, the potential is tremendous for leveraging capabilities which put advanced data science innovations into the hands of the masses. “By 2026, massive (>1 trillion parameter) foundation models for natural language processing , AI-generated images, and more, will become standard industry utilities provided only by the largest vendors,” Deepika Giri, associate vice president at IDC APAC, told Inside Retail. The biggest themes around ChatGPT revolve around areas of customer engagement, and the app could generate content for collaterals such as emails, marketing campaigns from web sources and ultimately manage customers with a high level of engagement. A brave new world “By 2026, 75 per cent of large enterprises will rely on AI-infused processes to enhance asset efficiency, streamline supply chains, and improve product quality across diverse and distributed environments,” she added. She said that IDC predicts that persistent talent gaps will drive 35 per cent of IT organisations to invest in AI skills by 2023, both to run automated IT operations and to support business end users adopting AI and automation solutions.   “In addition, there are legal, ethical and accuracy concerns surrounding AI adoption, which is largely unregulated today. We expect that there will be better laws and regulatory guidelines that will emerge to address these issues,” she stressed. At the end of the day, she is of the opinion that widespread adoption and trust in this space will depend on the level of governance that comes through, as would be the case with any new technology that emerges.