Whether it’s desk or cell phones, videoconferencing platforms or software-based trading systems, voice remains the primary communication tool for financial markets.
As a result, audio files and video recordings are an important part of the data that organizations review as compliance monitoring requirements around voice conversations have become stricter in recent years.
But capturing, monitoring and analyzing conversational data can be challenging and requires modern speech infrastructure to meet these requirements, says Nigel Cannings, founder and CTO at U.K.-based firm Intelligent Voice.
It’s even more challenging when it involves noisy environments like trading floors where traders speak fast and over each other, switch languages, use trading-specific jargon and where multiple parties can join or leave a conversation at any given time. That, and the ability to accurately analyze emotion and sentiments are some of the hurdles that financial service players face when reviewing audio and video files with limited human resources, he added.
“Humans leak a lot of emotion and data about themselves when they’re talking, whether it’s omitting certain types of pronouns or using distancing language to push themselves away from an event,” Cannings told PYMNTS in an interview. “And while we can train human beings to look out for these signs of deception, there’s a limited number of human beings available, and they can get very tired very quickly.”
It’s a gap that Intelligent Voice is trying to close with its artificial intelligence (AI)-powered speech and natural language processing (NLP) technologies, available in 24 languages and dialects, which he said can recognize behavioral cues and spot signs of frauds and suspicious activities, all in a matter of minutes.
Overall, Cannings said the goal is not to replace but rather complement human searching by alerting surveillance teams to potential fraud signs that may have slipped past them.
In fact, he said that leaving machines alone to make arbitrary decisions about customers is dangerous and leads to unconscious bias, as AI algorithms have the potential to be biased in ways that are still not fully understood today. “AI can discriminate based on your name, location and down to where you went to college, so it’s very important to have a human in the loop to help take a step back from it.”
Beyond the financial services space, Cannings said the advanced AI-driven speech technology offered by Intelligent Voice has enormous growth potential and can benefit governments in speeding up the rate at which genuine claimants get paid.
“One of the reasons why it takes so long for people to be paid benefits is because the government is inherently suspicious and is worried that people are making fraudulent claims” he said, pointing to the U.K. where billions of pounds were lost during the pandemic due to fraudulent claims.
Healthcare is another area where AI has the potential to revolutionize processes, Cannings further noted, especially since easy access to the internet has led to a spike in diagnoses that are often inaccurate and misleading.
“We’re overwhelming healthcare with unnecessary self-diagnoses,” he argued, explaining how AI can correctly analyze large amounts of data and enable patients to get genuine diagnoses that can then be funneled down the appropriate channels, saving them precious time on multiple visits.
“Now multiply that across an entire population and it almost becomes limitless,” Cannings said, adding that “anywhere where there’s a resource squeeze, AI and modern AI can help.”
For all PYMNTS EMEA coverage, subscribe to the daily EMEA Newsletter.