Chatbots. Personal assistants. Robo-advisors. Machine learning. Cognitive computing. And so much more. While the term artificial intelligence (AI) has been around for 60 years, it has finally become part of our daily lives—and how we bank, invest, and get insured. Some financial institutions have been investing in AI for years. Other firms are now beginning to catch up thanks to advances in big data, open-source software, cloud computing, and faster processing speeds.
A look back
Is it AI or is it not? AI means different things to different people. Here, we focus on what some call “weak AI”—machines capable of performing specific tasks that normally require human intelligence such as visual perception, speech recognition, decision-making, and language translation..
Something ventured, something gained. AI startups have raised more than US$2 billion in venture capital funding this year. This is clearly seen as one of the more promising technologies, with a bright future.
Not on the same page or algorithm. Each sector applies AI differently. For example, insurance leaders use AI in claims processing to streamline process flows and fight fraud. Banks use chatbots to improve customer experience. In asset and wealth management, AI adoption has been sporadic, but robo-advisors are rapidly changing that.
Going behind the scenes. Some firms use AI to model scenarios for capital planning, or use natural language processing and graph processing techniques to flag transactions for compliance reviews. These uses are lower profile, but they’ll have a big impact as they move toward mainstream.
Why aren’t more firms relying on machines? Two thirds of US financial services respondents said they’re limited by operations, regulations, budgets or resources limitations, according to our 2016 Global Data and Analytics Survey: Big Decisions
The road ahead
The machines won’t take over–yet. AI will gradually replace humans in some functions like personal assistants, digital labor, and machine learning. But challenges will persist because of bias, privacy, trust, lack of trained staff, and regulatory concerns. Augmented intelligence, in which machines assist humans, could be the near-term answer.
So much insight. With advances in big data, open-source software, cloud computing, and processing speeds, more firms will use cognitive computing and machine learning to perform advanced analysis of patterns or trends. For example, firms may use AI to help spot nonstandard behavior patterns when auditing financial transactions. Firms may also use AI to sift through and analyze thousands of pages of tax changes.
Make way for more robo-advisors. With the new DOL fiduciary rule, we could see an uptick in robo-advice due to pricing pressures on commissions. Robo-advice is also morphing to bionic advice by combining digital and human delivery of advice.
Cognification? Digital twins? What? We’ll evolve more from digitization and automation to what many now call cognification. We expect to see firms use AI more often with processes that rely on machines to make very specific decisions. We’ll also see companies modeling how customers might react to various scenarios, testing assumptions on users’ digital twins.
What to consider
Where to start? We recommend that you pick two different types of problems as you explore AI technology solutions:
- Some should be operational so that you show productivity improvements. Review and select the various AI technologies that can solve these problems.
- Others should be more exploratory in nature. For example, if you’re asking whether you can “get better customer satisfaction and retention by analyzing the audio data from call centers,” you might not have a specific metric in mind. However, applying AI to this problem may yield insight that other techniques can’t.
Make AI an extension of your data analytics team. Mature organizations might choose to set up a new chief AI officer role. But if your firm is in the early stages of adoption, view AI as an extension of current analytics capabilities instead.
Find the right balance between human and machine. There’s a balance between servicing costs and the need for good customer service. You should design off-ramps, swapping customers over to live support if an AI customer interaction or other transaction should falter.