I was recently in my kitchen asking my Google Assistant a simple question: “Hey Google. What’s the weather?”
The assistant responded, and I said, “Thank you!”
“You’re welcome, Andrew.”
I was pleasantly surprised. This was the first time the assistant had responded to me like this.
Interactions like this take one-directional commands to a new level, taking a great step toward making interactions with AI feel more human. But what happens when the command becomes more complex?
By Andrew Hamel
When you issue a simple command to a bot that understands language, whether a voicebot that understands speech or a messaging bot that understands text, you’re essentially asking it to respond to a single intent. But real conversations don’t work that way. In a real conversation that feels fully human, you often express multiple intents at the same time, add qualifying information, and share your emotion about how you feel about your problem or your options. Bringing in layers of information – like, “What’s the weather like at our destination over the weekend, and what do you think would fun to do there?” – makes things more intricate and interesting.
This type of request not only asks for the weather report, but assumes the AI remembers where you’re visiting. It also asks for a recommendation based on what’s known about the weather, your destination, and what it knows you like to do. The intents layered into the conversation have multiplied. This is how real conversations with other humans look, and it’s what conversations with voice and messaging AIs will evolve into.
We’re not quite there yet when it comes to the vast majority of consumer-facing bots. Many still exhibit some of these common pitfalls that we try to avoid when designing AI-powered experiences:
- The misunderstanding: The bot can’t discern what the user wants and gives its best guess. Close, but no cigar.
- Ignoring the question: The bot offers a solution that completely sidesteps the question. It’s confused and offers the only solution it knows.
- The loop: The user needs help logging into their account, but before the bot can help it asks them to log in. Hence, stuck in a loop.
- The botched transfer: “Hello? Is anyone there?” The bot fails to transfer to a human for advanced help, either because it doesn’t recognize when the situation calls for it or no humans are online.
- Frustration: Even when the bot does what it is supposed to do, sometimes the experience is frustrating because it can’t understand the emotion behind the ask.
Here’s the thing: AI is only as good as the data and integrations powering it. This is where big conversational data comes into play. It can help voice and messaging bots better understand natural language, and therefore help organizations that use AI to scale interactions sidestep the common pitfalls above.
For example, connecting conversational data across voice and messaging bots can give users a more seamless experience. No matter where they’re contacting you, they want to know that you remember them, their issues, and what they’re trying to solve. When bots have the context of the information they’ve previously shared or questions they’ve asked already, they don’t have to repeat themselves over and over again. Organizations can also create integrations that make it easier for automations to help users make appointments, sign documents, and make secure payments – all without ever leaving the conversation.
In general, it’s all about making things easier for the average user by making conversations feel more helpful and human at every touch point.
The average user is asking for this. Recent research surveying thousands of people shows that 88% of consumers prefer companies that connect the history of their interactions, and 82% prefer companies whose bots take what’s already known about them from previous interactions and apply it to their current situation.
Consumers even prefer using Conversational AI over human interactions for simple activities like booking appointments, updating addresses, and checking account balances. As these experiences start to feel more human, we can expect demand for AI-led consumer experiences to keep growing.
Empowering voice and messaging bots to speak to humans, well, like humans, is good for all of us. It frees up human agents for more complex and interesting tasks, plus cuts costs and boosts revenue for brands. Most importantly, it means we can expect more meaningful, personalized experiences as consumers. “You’re welcome” is just the start.
Andrew Hamel, EVP, Technology, Operations and AI, LivePerson
Andrew spent more than a decade working for Amazon leading development of the company’s recommendation engines, search function, and other machine learning-powered experiences. At LivePerson, Hamel leads the technology team creating the Conversational AI that helps brands provide better Curiously Human digital experiences — experiences where AI understands consumers’ intents, connects them to brands across messaging channels, and delivers meaningful outcomes for consumers, agents, and brands. Hamel is an expert in how big data is the key element to creating more natural interactions between customers and AI.
This article originally appeared in Inside Big Data.
The post Big Conversational Data Delivers More Human Engagements appeared first on TheCustomer.