By Rob McDougall, CEO, Upstream Works

In the rapidly evolving world of customer service, the role of artificial intelligence (AI) has taken on increasing visibility. Within the realm of contact centers, AI can serve as a powerful tool, offering solutions that range from agent-assisted service to self-service and a wide variety of other applications, such as translations, routing, and robotic process automation (RPA). However, the recent game-changer has been the introduction of Generative AI, with ChatGPT leading the way as its pioneering and most visible representative.

The Dynamic Duo: Assisted Service and Self-Service AI

When business leaders think about AI’s role in contact centers, two primary categories readily come to mind: assisted service, encompassing Intelligent Virtual Assistants (IVA) and Agent Assist tools; and self-service applications, embodied by chatbots. But as with any new technology, it becomes important to differentiate reality from hype. With AI, the landscape fragments into a diverse terrain of possibilities.

Generative AI: A Cost-Saving Marvel, but a Double-Edged Sword

Initially, businesses may view Generative AI, particularly when employed as chatbots, as a promising means to slash costs by replacing human resources with efficient bots. However, this simplistic perspective has many pitfalls.

Firstly, AI’s capacity to understand inquiries doesn’t guarantee its ability to take effective action. To illustrate, imagine asking someone to add themselves to your Customer Relationship Management (CRM) system or your billing system. While they understand your basic intention, they will lack essential details on how your business systems work, how to gain access, permissions, etc. Generative AI is no different. It has an excellent capability to understand your customer’s intention, but it needs to be connected and integrated into your backend systems to provide any business value to your customer. Integrations of this type – which have always been at the expense and limitation of self service, are, in the words of the band Talking Heads, ‘same as it ever was’. Providing these connections is a cornerstone of operationalizing any AI application.

The second challenge arises from AI’s propensity for hallucinations, creating unreliable responses. This inherent issue, intrinsic to Generative AI, necessitates a fundamental revaluation of neural networks and extensive retraining to rectify. Consequently, Generative AI is better suited for request categorization, gracefully bowing out while other systems furnish more standardized responses.

The third and most significant hurdle for Generative AI remains human nature; customers often harbor skepticism toward AI, irrespective of its accuracy or performance. This skepticism breeds lack of trust in the brand and leads to the customer’s ultimate reluctance to embrace AI-driven services.

The Balancing Act: Cost vs. Trust

Considering these complexities, implementing Generative AI chatbots incurs substantial setup, integration, and computational costs. Furthermore, it may provide unreliable information that erodes customer trust. Consequently, it is advisable to employ Generative AI for straightforward cases with minimal risk and uncomplicated responses, leaving complex interactions to human agents.

GenAI: The Agent’s Best Friend?

In contrast, when Generative AI serves as an Agent Assist, most of these challenges fade away. Putting the ‘human in the middle’ significantly lowers the risk of a bad AI experience for your customers. Integration becomes irrelevant, as the AI complements rather than replaces agents. Trust issues dissolve, as customers perceive agents as highly knowledgeable with AI assistance. However, the risk of hallucination persists, necessitating stringent input and output controls.

Streamlining Efficiency: Transcription and Summarization

Generative AI finds another valuable role in transcribing and summarizing interaction histories, a significant time-saver for agents. For instance, at Upstream Works, all chat transcripts, emails, and call recordings are tracked. Generative AI streamlines the process by summarizing this information across interaction histories, eliminating the need for agents to laboriously input summary notes. Further, when customers contact an agent, the interaction history is available in summary form to speed up the agent’s response capabilities. It truly gives the agent an edge – they can easily get up to speed on the customer’s history and provide personalized and contextualized service.

New Horizons: Mastering Translations

A final category reveals the versatility of Generative AI, with translations standing out as a remarkable advancement. Until recently, translations were marred by stilted and grammatically incorrect results, often provoking laughter from native language speakers. However, Generative AI has revolutionized translations, enabling seamless multilingual interactions. Now, an English-speaking agent can effortlessly engage in a Spanish chat with a customer, aided by AI’s language-switching prowess. In this scenario, AI operates discreetly, allowing companies to enhance the customer experience while economizing on multilingual agent teams and using larger, more efficient skill pools that are less language dependant.

Generative AI is really emerging as a transformative force in contact centers, offering diverse solutions that can be adapted to specific business needs. While challenges persist, the strategic deployment of Generative AI, whether as an agent assist, a chatbot or a translator, holds the key to unlocking enhanced customer experiences and operational efficiency. It’s a dynamic evolution that contact centers cannot afford to overlook.