“Give me a double cappuccino. Light sugar. Oh, make that two single cappuccinos. Decaf. Scratch that. Two singles and one double cappuccino. And make it all to go.”
Launched last week by the Alibaba DAMO Academy unit, the new technology lets customers at cafes and restaurants place their orders via voice with a smart-ordering machine and make as many changes or corrections as they want. The machine’s AI is able to catch all the changes and update the order in real time.
Alibaba Machine Intelligence Technologies addresses real-world problems, using AI applications to solve them. In this case, under the hood is breakthrough technology in spoken-language understanding, which involves both speech processing and natural-language processing. It required the research team to develop a way to capture both voice and visual features and take factors into consideration, like the speaker’s pace, pauses and breaths between words, pronunciation and facial expressions. As well, the team had to build a reinforcement-learning model that allows for revisions and intent detection.
“With the smart ordering machine as a perfect example, we believe the solution would greatly enhance the customer experience and make running a business more efficient,” said Yan Zhijie, head of Alibaba Machine Intelligence Technologies, noting that besides food and beverage ordering, the technology can be applied in customer service, voice commands for home appliances, in smart cars and voice response to inquiries in shopping malls and airports.
Machines featuring the new AI application will become available over the next few months, and the underlying technology solution will be available on Alibaba Cloud for the benefit of small and medium-sized enterprises. The technology currently only works for Chinese-speaking customers.
The smart-ordering solution follows on from the far-field voice-recognition technology Alibaba rolled out last December in Shanghai metro station ticketing kiosks. The first smart kiosk used in the Shanghai South Railway Station this year lets passengers use their voices to tell the kiosk where they want to go. The machines recommend the best route.
* Alizila is the independent (but Alibaba-funded) news resource on Alibaba Group activities.
