GenAI to Benefit Customer Service—Eventually

While generative artificial intelligence can help customer service leaders improve content creation, speed up data analysis, and make agent workflows more efficient, the technology is not yet ready for customer-facing use cases, Forrester Research asserted in a new report.

“Large language models aren’t the conversational miracle drug that contact centers are looking for,” Forrester cautioned, noting that there are too many significant issues, with too much risk, for customer-facing applications at this time.

Forrester said the highly touted technology could transform every domain within companies and called contact centers “a proving ground” for generative AI because of the massive amounts of unstructured natural language data that comes through them.

Though it expects generative AI to influence all elements of the contact center at some point in the future, the research firm sees the following as the largest impacts right now:

• Summarizing transcripts: Rather than having agents take minutes to summarize calls, generative AI can complete this task in an instant.
• Boosting productivity with quick replies to customer queries.
• Developing contextual content.

At the same time, though, Forrester cautioned that generative AI shouldn’t be approached as a silver bullet to solve all issues.

For one reason, all AI isn’t equal, according to Forrester. Large language models, it said, have been out longer than many people realize, but the hype only caught up late last year with the release of ChatGPT by OpenAI.

Since then, many vendors have launched what the research firm calls “half-baked, poorly tested products” in an effort to be first to market with solutions that incorporate generative AI.

“It’s essential that companies buying generative AI solutions carefully evaluate products to seek out clear differentiation, affirmations of quality, use case alignment, and security considerations. Don’t be afraid to ask vendors some hard questions, especially if they are claiming to generate unique content,” it recommended in the report.

Forrester also pointed out that agent workflow is one of the most relevant use cases for generative AI because the payoff could be high and the risk could be low. Yet the research firm cautioned against over-automating, which it said would make for worse agent experiences.

It’s important to balance acting on the agent’s behalf with helping agents make decisions, Forrester said, noting that while automating call summaries might seem like a good idea, it also eliminates the breaks that agents get between calls. Contact center leaders will need to fill agent time between calls with other value-added tasks so that their entire workday isn’t spent exclusively on the phone.

And finally, Forrester warned that generative AI will not solve for bad automation. “There aren’t enough controls in place to prevent harm to customers or a company’s reputation. Vendors are working on ways to combine techniques and create a semblance of guardrails to minimize these risks, but remember that there is more to a chatbot than the text that it generates,” it concluded in the report.