Generative AI in CX: A getting started guide
Go ahead: admit your secret use of generative artificial intelligence to David Truog. He’s becoming accustomed to the role of confessor.
Though his title is vice-president and principal analyst at market research firm Forrester, Truog’s advice has made him more of a trusted confidant for a wide range of customer experience (CX) leaders. As a result, they are being candid with him about their experiments with the most hyped technology in more than a decade.
“Just a few weeks ago, I was talking to somebody who, as part of her job as an executive in CX, needed to draft a letter to send to all their customers,” Truog told 360 Magazine. “She used her home computer and had ChatGPT draft it for her, and then she made a few edits and sent it out. She didn’t tell any of her colleagues that she used ChatGPT and nobody questioned it. That’s just the way it happens.”
CX leaders, like their counterparts in other business functions, are keeping their use of gen AI under wraps because in many cases it has not been officially endorsed by their company’s IT department, Truog said. He drew parallels with the early days of modern smartphones, when CIOs and their teams were resistant to any mobile device other than a BlackBerry.
Eventually, of course, IT found out how to secure and govern the use iPhones and Android devices, and Truog predicted the same will be true here. Eventually. “I think that’s IT departments across the country are dealing with figuring out how to not just say no, but how to do it safely. Which is a big shift.”
In the meantime, CX professionals are being bombarded with gen AI features or versions of the tools they commonly use for their work. This includes:
- Talkdesk, which has introduced features to ensure responsible gen AI in its contact center platform
- Ada, which has launched generative actions so chatbots can resolve issues without human intervention
- Invoca, which is offering conversation intelligence tools to train AI models more quickly
- Coveo, which is promoting its Relevance Generative Answering solution
- Cyara, which has integrated Open AI’s Chat GPT-3 for its conversational chatbot training and testing platform
Truog and his team at Forrester recently responded to the growing number of questions around these kinds of solutions and the best way forward with Generative AI Essentials For CX Leaders.
The report suggests CX leaders can use gen AI to build surveys and customer journey maps more quickly, gain greater insights from customer feedback and improve journey orchestration. It also warns, however, of three different privacy “traps” to avoid, the risk of bias against certain customer segments and possible unreliability in decision-making.
Despite the near-constant buzz about gen AI since earlier this year, Forrester is currently recommending gen AI should not be used for customer-facing use cases. There are exceptions, of course: Truog pointed to an experiment by 1-800 Flowers where customers could use gen AI to write poetry to accompany Mother’s Day orders.
“To use it to make up a fun little Limerick or something, that’s fine,” he said. “But for most business applications, the facts matter. Your life insurance company needs to ensure they’re offering customers the right answers when they get questions about their policies. You don’t want to be creative in that scenario. You want to have their backs.”
What to watch for in gen AI CX solutions
The Forrester report warns against CX vendors that offer “undifferentiated bolt-ons” based on the same widely-available APIs simply to keep up with the gen AI hype. It also notes the security vulnerabilities that the use of gen AI could introduce and the operational costs from more compute-intense workloads. That said, Truog said CX leaders shouldn’t necessarily wait on the sidelines for products to mature, any more than they should adopt solutions out of fear of falling behind.
“Find a middle ground: have a bot that doesn’t answer customers questions, but answers the questions of contact center agents, which then they can use in their interactions with customers,” he advised. “Because then you have a human in the loop who can use their judgment as a contact center professional and say, ‘Oh, that’s a really good, helpful answer.’
CX opportunities with gen AI
Besides helping improve the knowledge based contact center agents rely on to help customers solve problems, Truog said gen AI holds considerable promise in areas such as interaction design. Companies like Figma, for instance, are offering gen AI to make it easier to build digital assets like an e-commerce home page with a few simple prompts.
“People are not realizing what a big deal that’s going to be,” he said.
A big caveat to using gen AI in CX work is the long-term investment required. Truog said that many vendors are rolling out features free of charge for now. This means it will be critical to correlate the use of gen AI to metrics around productivity, efficiency and even areas like customer satisfaction and retention.
“Once (CX vendors) have gained a foothold, and their customers are using these gen AI solutions, it’s very likely that they’ll say ‘Okay, now we’ve figured out the usage and demand. We’re now going to come up with a real pricing model,” he said. “And most likely, it’s going to cost more certainly more than zero.”
