Here’s How the AI Assistants of the Future Will Become More Human

Since most consumer-facing artificial intelligence was built for commands, not two-way conversations, many of us are getting frustrated. Here’s how AI-driven experiences can get better.

Alan Gilchrest, Chief Technology Officer, LivePerson

July 1, 2022

4 Min Read
abstract group of chatbots
sdecoret via Adobe Stock

Alexa, stop!

Have you noticed that your in-home AI-powered assistants have gotten chattier lately? For many consumers, the love/hate relationship is tilting dangerously negative as their attempts to engage us fall flat.

We love these AI-powered experiences for sharing headlines and weather updates. And in general, they’re incredible showcases for technology built by some of the world’s best and brightest. But too often, our “conversations” with AI feel like efforts to upsell us or convince us to try novelty features we simply don’t care about.

It’s true that over the past few years, our collective comfort level with AI — whether found on voice-activated devices or via automated messaging with our favorite brands — has evolved considerably. In fact, positive customer sentiment toward AI-powered conversational automations, better known as "chatbots," nearly doubled from 31% in 2020 to 61% in 2021.

That’s good news for any organization using conversational AI to scale its reach. The tougher news is that the ever-growing volume of interactions with AI means consumer expectations are also growing every day. The data is clear: when consumers are asked about the weaknesses of conversational AI, one of their top concerns is the lack of a “human touch.”

As someone who worked on Alexa’s dialog management, I think what the average user is picking up on is the fact that most customer-facing AI experiences available today were built to take and respond to commands, not hold true two-way conversations.

But what if they went from simply responding to one-way questions to participating in a “more human” conversation? That type of engagement should be at the heart of any conversational AI experience. It’s what consumers want and expect out of AI-assisted devices in the home or on the phone or via messaging when contacting brands for support or sales.

To get there, we need to think deeply about what humans do well in conversations, and what really makes us feel heard and understood. Here are some of the most important things to consider as we build the more human AI experiences of the future:

1. Humans listen.

When you come to a person with a problem, they don’t (or shouldn’t) try to script their responses. The ideal state for conversational AI is to be able to respond to a conversation that unfolds naturally. To do that, they need to be built on Natural Language Processing (NLP). NLP turns language inputs, whether voice or text, into meaningful structured information like intent or product reference.

2. Humans resolve confusion and misunderstandings.

When you think about our interactions with each other, you quickly realize how much time we spend getting on the same page. Human agents are actually trained to restate your problem to you, so you know you’ve been understood. When it comes to AI-led conversations, the humans building them need to be able to peek under the hood to see when the AI gets it wrong. That means consumer-facing AI should be built on systems that allow you to easily identify points of failure, then drill down to fix them.

3. Humans adapt to cues.

When we talk about the “human touch,” we don’t mean AI that feels overly familiar and friendly. We want experiences that respond appropriately to how we’re feeling, connect with us, and help us work toward solutions. This means real-time sentiment understanding will only become more important as consumers continue to encounter AI as their first point of contact with organizations.

4. Humans hold multiple thoughts concurrently (and sometimes jump topics in conversations).

Conversations often jump from topic to topic, with simple requests going down different paths before returning to close out the original topic. AI must become more open to the flow of human thought rather than remain so structured and restrictive in achieving a conversational goal. Supporting multiple intents and reasoning engines to understand the most effective paths to resolution will be necessary to create more human-like AI.

5. Humans make connections to solve problems.
What good is a conversation if it can’t get you to the right person, place, or answer? Conversational AI can’t exist in a bubble. It has to connect to back-end systems, popular apps, and helpful services in order to make your day. For example, there’s a big difference between an AI that tells you that you need to set up an appointment and an AI that makes the appointment for you in real time. Integrating with the apps and services consumers use every day will make AI-led experiences feel more convenient and helpful than ever.

Making AI more human is a simple concept, but it requires advanced technology, analytics, personalization, and integrations. As we solve these problems, we’ll see a major shift toward AI experiences that have the human touch we’re all looking for.

Changing from command-based to conversation-based experiences will mark a new era for consumer-facing AI. When we get there, devices exemplifying the old ways of doing things will be put into glass boxes on display at the Smithsonian. We’ll remember them fondly as incredible pieces of technology that have been replaced by something even better.

About the Author(s)

Alan Gilchrest

Chief Technology Officer, LivePerson

Alan is responsible for building out the next generation of conversational commerce at LivePerson. He joined the LivePerson team as a Senior Vice President of Conversational AI in June 2018 in order to scale the company’s conversation technology through the novel approach of fusing conversational building blocks and AI-enabled conversational orchestration to redefine human-to-machine interactions. Previously, Alan spent more than eight years at Amazon working on Alexa as the Global Head of Dialog Management, AI Agent Integration and Next Action Selection. He helped create Alexa’s first data-driven dialog management framework leveraging deep insights from customer and environmental data.

Never Miss a Beat: Get a snapshot of the issues affecting the IT industry straight to your inbox.

You May Also Like

More Insights