Experts share 3 expectations about LLMs to set with business stakeholders

Since the launch of ChatGPT three months ago, large language models (LLMs) have rapidly transformed how people work across entire industries and disciplines. From content creation to customer service, conversational AI is well on its way to becoming embedded into our daily workflows.

After introducing AI Assist in January, Voiceflow brought together a panel of experts to discuss the implications of this trend: Greg Bennett (Salesforce), Daryl Thomas (Cisco), Kenny Tan (Intuit), Sibel Ciddi (U.S. Bank), and Rob Hayes (Voiceflow). In this blog post, we highlight their top three takeaways around how to set proper expectations around LLMs with business stakeholders. 

1. LLMs won’t steal your job

It’s important to communicate that ChatGPT and other LLMs can’t replace humans, the experts agreed. Otherwise, workers may panic and managers may plan for a future that never arrives.

The truth is, people have unique skills that machines can’t mimic, like the ability to empathize, imagine, and judge. We may be informed by the data in our environment, but we’re not dependent on it to function. That makes us inherently valuable to the companies we’re employed by. 

2. Teams will get more done, faster

While LLMs won’t replace workers, they will empower professionals to move faster and get more done. This is one benefit of embedding LLMs into your workflows that is worth highlighting, the panelists agreed.

By accelerating workflows, LLMs will restore employee bandwidth—empowering people to spend more time elevating the quality of their execution and improving the systems and processes that facilitate their activities. 

3. It’s not as simple as flipping a switch

Many executives are excited to invest in LLMs because they recognize their potential to drive positive business outcomes. But few leaders understand what conversational designers do behind the scenes to make LLMs useful.

Our panel of experts agreed that guardrails are necessary—a layer between what’s generated by LLMs and what’s delivered to the customer. And building rules, policy, and governance into conversational AI requires both careful consideration and cross-functional collaboration.

For these reasons, it’s important to set clear expectations around the need for human supervision when discussing LLMs.  

More on LLMs and conversation design

LLMs are now in Voiceflow. Read about these game-changing new features—and even more we plan to roll out soon.



LLMs won’t replace NLUs. Here’s why

No items found.