On 4-5 March, boost.ai was proud to attend the CX BFSI Exchange UK, a premier gathering of industry leaders, innovators and experts in the financial services sector. This exclusive event provided exciting insights and updates for professionals seeking to stay ahead in the rapidly evolving banking, financial services and insurance (BFSI) landscape.
Boost.ai Chief Technical Officer Rasmus Hauch was invited to speak as part of the panel session “Strategically Responding And Reframing Customer Experiences With AI To Deliver Satisfaction At Pace And Low Cost.” The panel sought to address many of the questions companies are increasingly asking about how they will use AI to add meaningful and long-lasting value to their customers and prepare for future trends.
Rasmus was joined by:
-
Matthew Harwood, Head of Customer Communications at NatWest Group
-
Ed Towers, Head of Advanced Analytics and Data Science Units at the Financial Conduct Authority
-
Lee Walker, Global Head of Extended Client Propositions, Digital Assets, Digital Security, Digital CX Barclays Corporate Banking
For anyone who was unable to be there, and for attendees who would like a recap, here are our top takeaways from this fascinating discussion:
How AI opportunities are enhancing the customer experience
For any enterprise seeking strategic insights on how to use AI to enhance the customer experience (CX), the greatest point of emphasis is to provide a highly detailed personalized experience that conveys a human touch. Customers are more responsive and loyal when they feel taken care of.
That personalization happens through customization. For example, in B2C situations, your representatives should be able to have a conversation with customers in which they can instantly reference previous conversations from across any and all platforms. Using AI assistance to pull up specific details from previous conversations can allow customer service personnel to efficiently guide customers, answer questions and resolve issues.
Other examples include:
-
Retail businesses using AI to track customers’ movements and activities within physical stores to understand how to better position products
-
Hotels using facial recognition to identify guests as they arrive, allowing staff to prepare their rooms and reservations ahead of time
-
Democratizing financial services by providing detailed financial insights for smaller financial institutions that don’t have the same resources as larger banks
In all of these CX situations, there’s a few considerations for how to effectively use AI models to help guide people:
-
Though all interactions should be highly personalized, it’s vital to create a standardized experience across all channels. That includes standardized branding and using a consistent voice and language.
-
Rather than trying to introduce completely new processes, AI should be used to support and enhance existing practices by implementing efficiencies and reducing waste.
Make full use of your resources by taking a hybrid approach in utilizing AI. How can you incorporate newer voicebots and virtual agents to your older, chat-based channels?
How we can ensure that data ethics and diverse perspectives are built into AI
As AI becomes more prevalent, there are increased questions about its ethical use as well as possible impacts of generative AI. It’s important to understand that this is an evolving area where it’s not easy to identify an answer. The definition of ethics or what constitutes diverse perspectives will largely depend on your industry, the country you’re operating in and your customer base.
Generally, ethical AI begins with careful consideration of the data you fed into the model and how you tested it. Today, we’re seeing increased emphasis on how to make models more representative of different users and how to influence the models used for generative AI.
For those of us who develop and utilize AI, there’s a few important guidelines that should always be front of mind:
-
An understanding that AI is only as good as the data it is trained on
-
An acknowledgement that humans can be biased in intentional and unintentional ways
-
A consideration of how AI development teams are structured
-
Establish guidelines, controls and functionalities over how to control the CX
-
Responsible AI development should have an eye on current and future laws and regulations
As your business expands into different countries around the world, there will always be specific guidelines and regulations — use them to help meet customer expectations.
What does the future hold for regulations on AI?
Another area of concern is the future of AI regulations. Different jurisdictions are taking different approaches, with the European Union taking an all-sector approach with regulation and proposing to ban the use of AI in certain contexts.
In contrast, the United Kingdom and United States are taking a more pro-innovation approach to regulation. AI regulation in those regions is shaping up to be more of a broad set of principles left up to individual regulators in varying sectors.
For example, in the financial sector, AI regulation is less about setting limits on AI itself but tying it directly to existing regulations related to financial outcomes and customer duty. When your financial institute is developing its AI strategy, you will have to ensure you are meeting compliance with defined regulatory elements such as security and resilience.
While we don’t know for certain how regulations will develop, we can share two important pieces of advice:
-
It’s vital for AI deployers to properly test their AI solutions to detect and prevent issues like hallucinations. There has to be self-imposed, tight control over all elements — from the data used to inform the models to consistent oversight over individual conversations.
-
For companies employing AI, understand that the most dangerous thing you can do is assume the AI will do everything and will require no oversight. That will inevitably end in some some sort of problem in terms of risk, whether it's reputational risk or concrete financial risk
We all need to play a part in balancing our compliance with regulations with the need to protect consumers.
How AI will impact upskilling or changing work roles
Will advances like generative AI fundamentally change the type or number of people that businesses need? The actual answer is still up in the air at the moment, though AI will undoubtedly have an impact on the workforce.
We’re at different stages of maturity with different use cases across many industries. The answer really depends on the skill levels required for different positions.
-
Creating new responsibilities. Responsible and effective use of AI necessitates the creation of new positions, such as keeping up the AI knowledge base and maintaining the high quality of its conversations. It’s important for every enterprise to have positions that understand how its AI processes work and where the guardrails should be.
-
Providing new tools. AI can allow for new capabilities with existing jobs. For example, in finance and banking AI can process billions of customer transactions every day. This is an extremely powerful tool for focusing on suspicious transactions to both stop fraud and reduce false positives. Enterprises will need to understand how to support those functions through better coding and better testing to deliver more with less.
-
Enhancing existing positions. Thanks to chatbots and virtual agents, many companies are seeing an increase in customers that are better able to reach out to the brands on a 24/7 basis. AI is also always available to assist employees in their work, and can help with difficult or complicated use cases. AI in a system improvement role allows for both better CX as well as greater employee satisfaction.
The expanding role of large language models in different industries
We’re seeing increased use of large language models (LLMs) in providing exciting synergies between voice and chat. As they evolve, LLMs allow for organizations to hold the same conversations in the chat as over voice with reduced friction.
We’re already seeing advances with LLMs that are able to capture more than just the written word. Expect more refined models to allow users to point a camera at an invoice and be able to relay information instantly. Or to incorporate videos into more than transcriptions, but also a detailed summary of the topic that allows readers to understand the essentials at a glance without having to read everything.
The main consideration for using LLMs and conversational AI is to train the AI so that it can have a uniform approach to its conversations with customers.
Final Key Takeaways
-
AI isn’t one thing, but rather a broad use of technology. While some AI platforms can be used relatively easily, such as simple programs built into major CRM solutions, there are more exciting and advanced tools on the horizon. Always take the time to consider what makes sense for your organization and focus on the areas that will make the most difference.
-
For AI deployers, it’s essential that testing and experimentation is done in a controlled environment. Make sure you can apply a specific control and set up the right criteria and metrics to define how specific conversations can be classified as “good” or “bad.”
-
AI has a lot of potential to update how we use money and conduct transactions. Many of our current financial processes are 40-50 years old and need to catch up to today’s digital economy. As we continue to push the envelope, it’s important to continually be aware of current and developing regulations and obligations to customers.
When considering the use of AI, focus on the outcomes you’re trying to achieve and whether it will be enhanced by employing different AI platforms.
Boost.ai is a leader in conversational AI platforms for enterprise organizations. We help businesses transform their customer and employee experience through the use of AI-powered virtual agents. If you’re interested in maximizing the potential of your conversational AI tools and want to start your conversational AI journey, connect with one of our experts today!
