BY RENÉ CLAYTON
Will there be a point in the near future where technology becomes so proficient that it overtakes all good? Pervasive in its growth, we may not even notice the areas where artificial intelligence (AI) is currently at play, but we are seeing that it has a variety of implications for those that choose to develop and deploy AI systems.
During the recent American Banker’s Digital Banking conference, several thought leaders – myself included – weighed in on implications regarding recent AI developments in the financial services industry. In particular, we focused on how the spread of AI might affect credit unions across the United States.
The takeaway? Speech analytics, facial recognition technology and natural-language processing chatbots contribute to marketable advancements within the credit union industry, but they do not exist without careful considerations, security awareness and potential liability risks. The engines that run these technologies rely on the quality of data with which they are continuously fed. If the data sets are biased, AI responses will contain and reflect those biases.
AI is everywhere. It has been incorporated into smart speakers, Netflix accounts, payment chatbots, smartphones and wearables – all of which produce data. Intelligent machines develop intellectual predispositions based on what they have been exposed to and mitigating these risks of exposure is more important than ever.
Since most of us are not wishing for the possibility of irrelevance, there is some good news: human involvement remains essential when it comes to AI. There are a few ways humans can and must stay involved in the development and deployment of new AI technologies for the purposes of quality maintenance and compliance. For example, humans are always expected to test AI assumptions by performing proof of concepts on a variety of experimental hypotheses. They must also determine what worked, what fell short and what cost too much, even if the outcome was successful. Lastly, humans need to develop a plan around the results of the above analysis to ensure that all elements are addressed as necessary.
On a similar note, there are a few best practices for AI development to consider, like initiating an enterprise strategy which has executive support and determining where the responsibility lies within the organization. Creating governance and security practices around the AI toolsets, as well as keeping legal, regulatory and compliance implications at the forefront of your planning, is also recommended.
Be cognizant of challenges, such as access to clean and categorized data and integration with existing systems. Avoid an over-emphasis on rules-based programming versus natural philosophy. Accuracy with natural language processing is also key. In fact, according to VentureBeat and Retale, 53 percent of those who had interacted with chatbots said that they needed to be more “accurate.” Talent acquisition of AI data scientists and business alignment within the organization need to be addressed from the outset of an AI project. Finally, don’t forget to provide a fair estimation of the time and effort it takes to get an AI-powered system up and running.
As the rise of AI and interconnectedness continues, it is essential that credit unions and other companies ensure they have protected consumers and themselves from the implications. Real human intelligence is not one-dimensional. If financial services companies are to capture the true depth that characterizes human reasoning, we must stake a claim in the development process and run repetitively through safety checklists. No matter how advanced our technology becomes, we are still responsible for the outcome. People will continue to ask who is handling, seeing and managing their personal and financial data, and the answer to this question will come more easily to those who take a proactive role in the process of creation and curation of AI.
As an innovation strategist, René Clayton provides a strategic vision for multiple solutions across PSCU’s organization, including the digital and fraud lines of business. She has more than 18 years of experience in technology and product development. René is a champion for AI and is passionate about new ways of creating amazing digital experiences. Recently, she led the winning team in an Internet of Things (IoT) project in the “Easiest to Explain to Mom” category during the PYMNTS.com 2018 Voice Challenge with Amazon Alexa. In 2017 and 2018, she successfully completed two courses focused on AI and IoT implications on business strategy through MIT’s Management Executive Education program.