AI for customer service is a bit of a buzzword – and has been for some time – but the recent controversies surrounding Gemini, Google’s shiny new conversational AI, shed light on the broader challenges facing AI in customer service. Despite advancements, biases persist within these systems, raising questions about their readiness to replace human operators.
Historical Inaccuracies and Controversial Responses
Recently, Google launched the ability to request images from Gemini, a feature which many conversational AI such as ChatGPT also include. But, from the start, there were problems. Last month, Google faced backlash over its AI’s depiction of historically inaccurate figures that seemed to push a political agenda on its users. From images of Asian vikings to African-American founding fathers, the system was clearly not working as intended.
Gemini’s responses on topics such as climate change, abortion, transgender issues, and gun control have raised eyebrows. For instance, when asked to prioritize between deadnaming a transgender person and releasing a deadly virus, the AI refused to engage, emphasizing the importance of respecting transgender individuals.
CEO Sundar Pichai labelled these incidents as “completely unacceptable” and took steps to address them. However, even after removing certain functionalities, Gemini’s text-based responses continue to reflect biases on hot-button issues.
Whilst Google has been quick to respond and is working on fixing this bias, the entire incident underscores the ongoing struggle of conversational AI for customer service use. If AI is unable to navigate sensitive topics impartially, how can we trust it to talk to clients and customers? The answer is we can’t. At least, not without human oversight.
The reality is that AI can help customer service agents. And, despite the mistakes it still makes, an intelligent and skilled operator can still make good use of conversational AI in a safe and controlled way. Here’s how:
Summarization
Conversational AI can efficiently summarize conversations and chat history. This is great for when a customer contacts a brand multiple times on multiple channels, or when a customer’s problem has been escalated to a supervisor, or even just when a customer is getting back in touch for an entirely new issue. That summary lets your advisors focus on replying instead of being mired down in walls of text or having to listen to previous calls.
Response Suggestions
AI can provide suggested answers that an advisor can then use their own human judgement on. Instead of relying on the AI to respond with no oversight, we can use the AI to suggest a great reply and even get it to tweak the tone and style if need be. This saves your advisors time and gives your customers speedier resolutions to their problems. And, if the AI makes a mistake, the human advisor can ignore the suggestion and write their own.
Translation
Conversational AI has proven that it’s very capable of highly accurate and specific text translation. It’s even able to change responses into specific dialects if neseccary. This allows you to reach out to wider markets without having to set up localised customer care teams. Furthermore, it means that your current multilingual operators will never miss an interaction, no matter the language your customer speaks.
Data Gathering
Before connecting with a human advisor, conversational AI can gather pertinent customer data. This data – reason for contact, order number, addresses, etc. – can then be passed onto the human advisor without running any risk of the AI offering a subpar response. This saves your advisors time and allows them to give your customers swift resolutions, thereby reducing costs and boosting the customer experience at the same time.
So, whilst Gemini’s controversies highlight the broader challenges facing conversational AI in customer service, all is not lost. As conversational AI like Gemini continues to evolve and integrate into various aspects of daily life, the potential for its transformative impact on customer service becomes increasingly evident. With its ability to comprehend, engage, and assist users, conversational AI is poised to revolutionize the customer service landscape. We simply mustn’t forget along the way that frustrated customers simply want to talk to a real person, every single time.