At the API World / AI Dev World virtual conference held this week, Joe Bradley, chief scientist with LivePerson, a developer of commerce and AI software, gave a keynote on ways AI and natural language software can change some dynamics of commerce and consumerism. Connecting and understanding customers can be more meaningful than offering a convenience, he said, and could help decide the winners and losers among brands as commerce evolves further.

AI and automation are intrinsic to natural language, conversational systems many brands deploy to speak with their customers, and while this might free up employees to focus on other tasks, there are some recommendations Bradley offered. “If you’re a builder of a conversational system, make sure you know how to measure conversation quality,” he said. “That’s highly difficult to do. It’s not a deep enough field of research today. There really isn’t an industry standard.”

Commerce is undergoing a transition that can be tumultuous, Bradley said, where consumers have changed how they digest digital media and shop. “For the brands that figure out how to connect to consumers best, this is this kind of a time for massive adoption and re-adoption.”

He also spoke of the radical change brought to commerce with the advent of Sears in the late 1880s with its catalog, which industrialized and reshaped interactions for consumers. Much like those catalogs, improving the consumer experience through AI and conversational communication software in ways that show an understanding of the customer, can help brands build trust, Bradley said. “Interactions where we’re better understood often feel quite different from the sleek, modern picture we have of commerce where everything is neat, tidy, and almost impersonal,” he said.

There are times when technology can act as a barrier to brand interactions. When customers want to talk to a person at a brand to solve a problem or find more information, Bradley said they tend to hit a so-called digital fortress. This can include waiting on hold on the phone or chatting on a website with a representative to resolve an issue. and then run into roadblocks. “A lot of companies have built this over time out of their customer service interactions,” he said. “The bigger problem here is we’ve allowed companies to industrialize and separate how they think about customer service from how they think growth and marketing.”

There might be poorly measured, poorly understood business processes that are not connected to a holistic view of the customer, Bradley said, and that drive the propagation of the digital fortress. 

“That’s a real weakness of modern ecommerce,” he said. “You tend to see it more in companies where there’s a big barrier to entry for their competitors.” That affords such companies leeway to offer customer experiences that are tough to navigate because they have less risk of losing business to a rival compared with much more commoditized products, Bradley said.

Commoditized products face different issues connecting with customers, he said, where the customer experiences that brands create become very similar, if not indistinguishable.

Conversational experiences can help brands understand their customers better and potentially establish trust, Bradley said. “There’s a huge opportunity to connect a little deeper, but there’s a scale problem. You want to automate as many conversations as you can. You want to use computers to do this and reach as many customers as possible.”

Bradley highlighted the importance of getting natural language understanding right when building machine-based conversational agents such as AI or chatbots. “If you’re building these systems, really be thinking about investing in your natural language understanding because that’s what creates cognitive points,” he said.

Unlocking natural language applications further might open new aspects of convenience in digital shopping, Bradley said. That could mean consumers keep speaking in their own manner to digital resources, expressing what they want rather than thinking about how to phrase thoughts based on the requirements of a search engine or machine. “Brands and companies building systems more like this are taking cognitive load off the customers,” he said. “Not nearly enough conversational systems today work like this or close enough to this.”

Brands should not assume they will build a conversational system once and be done, Bradley said. Rather they should treat it like a product they will improve over time, which will include keeping human beings in the loop. “Computers can’t solve every problem today and you should tie the situation to real business outcomes,” he said.

Related Content:

CIO Agenda: Cloud, Cybersecurity, and AI Investments Ahead

How AI Will Shape the Post-Pandemic Office Experience

Managing Uncertainty With AI and Big Data Beyond the Pandemic