fbpx

Alibaba’s Voice Assistant Easily Outshines Google

Alibaba AI sweeps Google under the rug

Alibaba’s Voice Assistant Easily Outshines Google

Think Google has the voice assistant game hacked? Think again.

Sure, Google’s Duplex solution with its creepy human-like resonance made a few waves when it appeared back in 2018 – but now that’s old news.

It’s 2019, and there’s a new giant on the scene – one that’s been quietly clocking in overtime while Google’s engineers sleep. We’re talking about Alibaba’s very own voice assistant, Tmall Genie.

When Duplex appeared in 2018, we were all impressed by its natural-language understanding capabilities and insights into the future of AI-human interaction. But Alibaba might have been the tortoise beating the hare, and we didn’t even notice!

On December 2nd, Alibaba demoed their new AI customer service agent for logistics company Cainiao at the Neural Information Processing Systems conference, and already, the agent is handling millions of customer requests.

What Makes This AI So Special?

So, why is Alibaba’s voice assistant so impressive compared to options like Google Duplex?

Well, for one thing, this AI gives customers a natural and conversational experience, without also giving them a case of the heebie-jeebies. As part of their December demonstration, Alibaba showed the assistant handling conversational elements in a conversation with ease. The agent paused when the customer interrupted it, repeated useful information, and responded easily to nonlinear conversation.

These are the areas where Alibaba’s agent truly shines. First, the agent can respond to interruptions and relay useful information without having to start from scratch when something doesn’t go “according to script.” Secondly, when non-linear conversation occurs, the agent doesn’t lose track of its purpose, it answers customer’s questions, then goes right back to the initial script. Thirdly, Alibaba’s assistant can respond to “implicit” intent. In other words, it can infer basic context in a conversation.

That’s way more revolutionary than a bot like Google Duplex that leaves customers with visions of Terminator-style global takeovers.

Human Conversations without the Horror

While some early testers were horrified by Google’s approach to making conversations more “human”, Alibaba’s voice assistant has taken a different approach by addressing the human elements in discussions that bots previously struggled to handle.

Of course, up until now, we’ve only seen what the Alibaba bot can do on stage, which means that there’s always a chance that the agent won’t respond as well in a day-to-day environment. For now, the assistant is only being used to manage package deliveries, but the Dean of the Machine Intelligence and Tech lab for Alibaba, Jin Rong says that other options could be in the pipeline.

When it comes to figuring out how Alibaba’s assistant got so smart, Rong wouldn’t share his secrets, but he did suggest that the thousands of recordings that the company has at its disposal are a valuable asset. On an average day, Alibaba handles around 50,000 service calls.

What’s Next for Alibaba?

The biggest advantage that Alibaba has when it comes to creating new AI solutions is the huge wealth of data it has access to. Machine learning relies on information, and Alibaba has that by the bucket-load.

Currently, the company is in the process of developing additional digital assistants to support other parts of its business, including an agent that will be able to take orders in noisy restaurants throughout the food industry, and a human-like virtual avatar that can answer questions about products. There’s even a chatbot that can haggle over prices with customers on the resale platform, Xianyu. We definitely haven’t seen that from Google yet!

Each of the assistants available from Alibaba is defined by the incredible speech recognition software and natural language processing solution, AliMe. We’re interested to see where Alibaba goes next.

 

Got a comment?

Please login to comment

Login

Popular Posts

Related Articles