AI is rapidly transforming the customer experience (CX). New tools are helping businesses drive better efficiency, personalize experiences and optimize workflows. However, deploying the technology, especially for customer-facing interactions, requires a careful balance between innovation and safety.
Manish Pandya, Senior Vice President of Digital at TaskUs and Dr. Paul Dongha, Head of Responsible AI and AI Strategy at NatWest Group talked through real-world examples and best practices for securing and improving AI models in a webinar, “The Impact of AI on Customer Service,” with AI Business.
Here are three important pieces of guidance from the discussion.
‘Have a customer improvement focus via continuous feedback loop.’
GenAI is non-deterministic, which means that large language models (LLMs) generate different responses based on how and when a question is asked. To ensure the accuracy of AI outputs and a positive digital customer experience, Manish says, “Look at all the interactions, rate them and use them to iterate and improve your application.”
He also points out that humans not only facilitate customer interactions but also play a crucial role in refining a model’s output and enhancing its performance.
‘Responsible AI and ethical risk management is not just about putting in a system.’
AI comes with potential risks: data breaches, intellectual property (IP) infringement and hallucinations or where a model generates misleading information.
To mitigate these, Dr. Dongha emphasizes, “Approaching AI responsibly is about people and governance, processes and technology.” Businesses must implement guard rails to monitor and validate prompts and responses. They can also establish an ethics oversight board that will evaluate use cases and ensure deployment aligns with norms and brand values.
‘All technologies provide an outcome that should be measurable.’
“You need to have a success program,” according to Manish. Tracking improvements in KPIs, profit margins, overall profitability and operational footprint helps measure the effectiveness of a tool and ROI.
He also highlights that an iterative approach with GenAI enables organizations to evaluate whether the operational efficiency gains are real and contribute to improved business outcomes.
