Today, most banks, insurance companies, and other types of financial services businesses have deployed natural language processing (NLP) tools to meet some of their customer service needs. But most of these tools fall far short of the organization’s goals for technology.
In many cases, these financial services companies could close 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 companies around the world are investing heavily in artificial intelligence (AI). According to IDC, global AI spending is likely to reach $204 billion by 2025. Banking is the second biggest spender, with most of that investment going to NLP-powered automated customer service agents, or chatbots. . Juniper Research predicts that chatbots will account for 79% of successful mobile banking interactions in 2023.
But while financial services companies 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 their chatbots can do this today. Similarly, 91% of respondents wanted their chatbots to automate actions based on customer responses, but only 52% said their current technology had this capability.
While NLP is undeniably useful with its ability to compute words and text, the complexity of human language presents serious challenges. NLP-powered chatbots often struggle to grasp 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?
Chatbots used by financial services institutions are conversational interfaces that allow humans to interact with computers by speaking or typing normal human language. Some of them use NLP technology while many are simple rule-based interfaces that follow a prescribed flow without any AI.
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 best conversational AI systems are about to pass the Turing test, that is, they are very difficult to tell apart from a human being.
A few very advanced chatbots powered by conversational AI will allow customers to ask more complicated questions. For example, 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, perform analysis on your purchases and give you a total. If you pursue this question by saying, “What about Dubai?” the conversational AI would understand from the previous context that you were asking how much you spent.
Good for customers, good for business
Customers find conversational AI much less frustrating than other types of chatbots. Due to their advanced NLP capabilities, these tools are much more likely to understand customer needs and deliver the appropriate service, in the required regional language and dialect. It can also help speed up customer service interactions and provide sophisticated support any time of the day.
And while many companies are deploying chatbots to reduce face-to-face interactions with customers, researchers say those powered by conversational AI tend to increase customer engagement. But that’s not a bad thing. Engaged customers tend to purchase additional products or services and become even more loyal customers.
Investments pay off in more than increased customer loyalty. Juniper Research predicts that by 2023, global operational cost savings for chatbots in banking will reach $7.3 billion, and for insurance, savings will reach $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 this not only benefits customers, but it can also boost employee morale.
Conversational AI also collects loads of useful customer data. Conversational AI offers better insight into customer intentions and emotions than other types of chatbots or even human beings. And because the conversation is already digital, it doesn’t need to be recorded and transcribed before it’s available for analysis.
Common challenges with conversational AI
These benefits make the technology extremely attractive to financial services firms. But before launching a new conversational AI project, be aware that deploying these chatbots also comes with some challenges.
As with all financial services technology, protecting customer data is extremely important. In some parts of the world, companies are required to host conversational AI applications and store associated data on self-managed servers rather than subscribing to a cloud-based service. Data integration can also be difficult and should be planned from the beginning of the project.
NLP technologies must be carefully trained and extensively tested to ensure that they have no biases. This hard work pays off when the tool can effectively connect with a wider audience without excluding or offending anyone.
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 companies host on-premises and need to research 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 application scope and throughput requirements. Some conversational AI implementations rely heavily on ML tools that incorporate neural networks and deep learning techniques. Many of these more advanced chatbots work 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 does not incorporate ML techniques. These companies are seeing great performance with superior ROI on Dell Validated Designs for AI. These systems also have the advantage of being modular to support rapid scaling as your chatbot usage grows.
Read Dell Technologies’ Conversational AI white paper to learn more.
***
Intel® Technologies Advance Analytics
Data analytics is the key to unlocking the maximum value you can extract from your organization’s data. To create a productive, cost-effective scanning strategy that gets results, you need high-performance hardware that’s optimized to work with the software you’re using.
Modern data analytics covers a range of technologies, from dedicated analytics platforms and databases to deep learning and artificial intelligence (AI). New to 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 broad ecosystem of analytics technologies and partners, Intel accelerates the efforts of data scientists, analysts, and developers across industries. Learn more about Intel Advanced Analytics.