Today, most banks, insurance companies, and other kinds of financial services firms have deployed natural language processing (NLP) tools to address some of their customer service needs. But most of these tools fall far short of organization’s goals for the technology.
In many cases, these financial services firms could fill in the gap between their expectations and their current capabilities by deploying a chatbot with conversational AI capabilities.
The rise of chatbots… and their weaknesses
Financial services firms all over the globe are investing heavily in artificial intelligence (AI). According to IDC, worldwide spending on AI will likely top $204 billion by 2025. The banking industry is the second biggest spender, with the largest portion of that investment going toward automated customer service agents powered by NLP, or chatbots. Juniper Research predicts that chatbots will account for 79% of successful mobile banking interactions in 2023.
But while financial services firms recognize that chatbots are the future, there are significant challenges. A Forrester report commissioned by vendor ADA found that 95% of financial firms would like their chatbots to understand customer history with the company. However, only 55% said that their chatbots could do that today. Similarly, 91% of respondents wanted their chatbots to automate actions based on customer responses, but only 52% said their current technology had that capability.
Although NLP is undeniably useful with its ability to compute words and text, the complexity of human language presents serious challenges. Chatbots powered by NLP often have a hard time capturing the context of words in a sentence, cannot detect sarcasm or tones of voice, and get stuck on words with multiple meanings.
How is conversational AI different?
The chatbots used by financial services institutions are conversational interfaces that allow human beings to interact with computers by speaking or typing a normal human language. Some of them use NLP technology while many are simple rules-based interfaces that follow a prescribed flow without any AI at all.
Conversational AI is a highly advanced application of NLP that allows human beings to have a spoken or written conversation with a computer system. The very best conversational AI systems come close to passing the Turing test, that is, they are very difficult to distinguish from a human being.
A few highly advanced chatbots powered by conversational AI will allow customers to ask more complicated questions. For instance, they might be able to ask, “How much did I spend in Paris last month?” And the chatbot would be able to understand what you were asking, run analytics on your purchases, and give you a total. If you followed up that question by saying, “And what about in Dubai?” conversational AI would understand from the prior context that you were asking how much you spent.
Good for customers, good for companies
Customers find conversational AI far less frustrating than other kinds of chatbots. Because of their advanced NLP capabilities, these tools are much more likely to understand what customers need and provide the appropriate service, in whatever language and regional dialect necessary. It can also help speed up customer service interactions and provide sophisticated support any time of day.
And while many firms deploy chatbots to decrease face to face interactions with customers, researchers say that those powered by conversational AI tend to increase customer engagement. But that isn’t a bad thing. Engaged customers tend to buy additional products or services and become even more loyal customers.
The investments are paying off in more than increased customer loyalty. Juniper Research forecasts that in 2023 the global operational cost savings from chatbots in banking will reach $7.3 billion, and for insurance, the savings will approach $1.3 billion.
But these monetary savings, while significant, are often less important in the long run than the time savings. By handling most low-level tasks, conversational AI can free up staff for other activities. And that not only benefits customers, but it can also increase morale among the employees.
Conversational AI also collects heaps of useful customer data. Conversational AI provides greater insight into customers’ intentions and emotions than other kinds of chatbots or even human beings can provide. And because the conversation is already digital, it doesn’t need to be recorded and transcribed before becoming available for analysis.
Common challenges with conversational AI
These benefits make the technology extremely attractive to financial services firms. But before kicking off a new conversational AI project, be aware that deploying these chatbots also comes with some challenges.
As with all financial services technologies, protecting customer data is extremely important. In some parts of the world, companies are required to host conversational AI applications and store the related data on self-managed servers rather than subscribing to a cloud-based service. Data integration can also be challenging and should be planned for early in the project.
NLP technologies need to be thoughtfully trained and tested thoroughly to ensure they don’t have any biases. This hard work pays off when the tool can effectively connect with a wider audience without excluding or offending someone.
Infrastructure designed for conversational AI
Conversational AI can be hosted in a public cloud service or in a company’s data center for control, compliance and security reasons. Many financial services firms host on-site and should investigate what kind of hardware is needed and whether potential vendors have systems designed specifically for conversational AI.
So, what kind of hardware is needed for a conversational AI application?
The answer depends on the scope of the application and throughput needs. Some conversational AI implementations rely heavily on ML tools that incorporate neural networks and deep learning techniques. Many of these more advanced chatbots run best on high-performance computing (HPC) clusters with dozens of Dell Technologies PowerEdge server nodes, NVIDIA GPUs, and fast storage.
Other organizations choose to deploy conversational AI that is more limited in scope — perhaps it supports text-only rather than voice and doesn’t incorporate ML techniques. These firms achieve excellent performance with superior ROI on Dell Validated Designs for AI. These systems also have the advantage of being modular to support rapid scaling as usage of your chatbot increases.
Read the conversational AI whitepaper from Dell Technologies to learn more.
***
Intel® Technologies Move Analytics Forward
Data analytics is the key to unlocking the most value you can extract from data across your organization. To create a productive, cost-effective analytics strategy that gets results, you need high performance hardware that’s optimized to work with the software you use.
Modern data analytics spans a range of technologies, from dedicated analytics platforms and databases to deep learning and artificial intelligence (AI). Just starting out with analytics? Ready to evolve your analytics strategy or improve your data quality? There’s always room to grow, and Intel is ready to help. With a deep ecosystem of analytics technologies and partners, Intel accelerates the efforts of data scientists, analysts, and developers in every industry. Find out more about Intel advanced analytics.
