Bank AI 2017 Conference Review (Part 1 of 2)

by on Oct 23, 2017

Artificial intelligence has been the biggest buzz word in the financial services industry in 2017. Though several prominent banks have already unveiled rudimentary chatbots on popular messaging platforms like Facebook, and many more plan to follow suit, we have yet to see a wide range of AI applications come to market. In that sense, 2018 may be a turning point for AI in banking, with more advanced AI-powered tools slated to become available. Given the unrelenting buzz over AI this year and the air of anticipation for 2018, thought leaders and tech innovators gathered in Boston from October 16-17 for the inaugural Bank AI 2017 conference. Corporate Insight attended the event, which sought to highlight where AI stands today with respect to financial services and what advancements we can expect to see in the coming years. Below, we present takeaways and highlights from the conference.

Conference Takeaways & Highlights

1. Clean Your Data! – AI algorithms are only as good as the underlying data they use, an idea discussed by multiple presenters and panelists. For years, banks have been collecting data on their clients without clear plans for what to do with it. Only recently have firms begun scrubbing the data and meta-tagging and categorizing it with any consistency. In order for an AI program to be successful, it must have access to viable and accurate data. Several banks that were quick to jump on the AI train found there were huge data-cleansing roadblocks that stood in the way of their initiatives. On the AI-adoption roadmap, many firms are still in the process of taking unstructured bank data and making it usable. One of the main challenges in this process is figuring out which data is actually needed and categorizing it effectively.

2. Use Your Chatbots Wisely – 2017 saw a mad dash to launch client-facing chatbots. There are several motivations to offering chatbots, including reducing back-office costs for the firm by having fewer CSR interactions, reaching a broader customer segment 24/7 and even cultivating the image of a tech innovator. That said, firms must carefully evaluate the role of a bot as part of their service offering rather than impulsively jumping on a popular trend. Currently, most active bank bots largely duplicate existing functionality that is adequately offered through other digital channels (e.g., looking up balance and transaction data), providing minimal incremental value for clients. As a smarter approach, bots could be leveraged to improve tasks that are inefficient or flawed. They can reduce call handle time and call center efficiency by determining customers’ needs and addressing their most pertinent issues before passing them on to human CSRs for more involved help.

3. Rise of the Business Translator – To launch any AI initiative, firms must have a strategic plan that makes business sense, a task made difficult by the inherent complexity of AI and its underlying algorithms. Often, these algorithms are created by data scientists who understand advanced equations and concepts, but lack business know-how. As such, the role of a “Business Translator” has become increasingly important. The idea behind this position is to have someone who understands the data (if not the underlying algorithms) and can translate it to real-world problems and solutions. A successful translator can ask the right questions and come up with the business use cases. This can be difficult to outsource and often requires access to internal or proprietary knowledge.

4. Democratization of AI – At this stage, AI is in the domain of the people who develop the models, but soon we might see tools come to market that make it easier for the wider population to build AI models. The first banks that launch effective AI platforms will have the initial advantage, though likely short-lived. Eventually all firms will use AI in some capacity, whether they build it in-house or source it through third-party companies and partnerships. As AI is refined and more options become available, the playing field should quickly level off. Most likely, the most important models will still be developed by data scientists with PhDs, but it is increasingly likely that engineers will soon be able to create simpler or basic AI programs that meet their needs. Companies like Microsoft already offer a portal of plug-and-play AI solutions that can be quickly implemented by financial services firms.