Forget virtual assistants and chatbots, the future of artificial intelligence (AI) and machine learning is seeing into the future.
In recent years we have begun to see how AI applications have established themselves as a staple in business. Alongside cloud computing, big data and mobile internet, AI solutions have found a preliminary use cutting costs on basic interactions and processes.
But their real value going forward is joining the dots between the masses of data points, interactions and captured behaviors to provide accurate predictive analytics. We can’t use a crystal ball to see into the future — but increasingly, we can use AI. Let’s have a look at some of the sectors in which AI is helping businesses see beyond tomorrow.
Banking
AI has been making solid strides into banking, aided by the forward thrust of app-based challenger banks and the demands of increasingly mobile consumers. The need for a frictionless mobile experience with a low cost base has meant AI has been able to make inroads across various different areas of the banking world.
In customer service, for example, sophisticated chatbots and virtual assistants are increasingly on hand to parse natural language requests from customers and automate basic processes, such as account opening and closure, or password or login changes.
They can also take some of the hard work out of handling more complex queries, by taking care of preliminary information capture with the customer, before handing off to skilled personnel.
This not only saves time for the customer, resulting in fewer lost sales or abandoned website visits, but can help banks to focus human resource where it is actually needed, with a dramatic effect on cost.
Those same AI skills, however, can also work hard behind the scenes, scanning large amounts of text, such as complex legal documents, and pulling out the key pieces of information, in a fraction of the tile it would have taken trained personnel.
The future of AI in banking, however, will involve a far deeper involvement in risk protection. By effectively joining up the dots between transactional data and processes, such as electronic identification and verification and know your customer, machine learning can be used to pinpoint high-risk customers, who may need to be screened with Enhanced Due Diligence (EDD) processes, or even predict cyber fraud and head it off at source.
Law enforcement
Of course, any system that can analyze human behavior and forecast fraud is going to appeal to law enforcement agencies. That is exactly what has happened in the UK, according to a piece first published in New Scientist.
West Midlands Police, a British regional force, is reportedly leading the trial of prototype “pre-crime” software called National Data Analytics Solution (NDAS). Driven by recent cuts to funding, the cash-strapped Force is hoping to save money by preventing crime rather than solving it.
The software sends its artificial intelligence algorithms to work with footage from smart cameras around cities and crunch crime statistics to predict the likelihood of someone becoming a victim or a perpetrator of crime — and then intervene.
If that sounds far-fetched, then it’s worth bearing in mind that the system can only work with information that is already out there. In the trial, West Midlands Police mined existing data and statistics from past criminal events to identify some 1,400 potential indicators for crime. The NDAS algorithms then took the data points and learned how to detect crime while analyzing video from metropolitan smart cameras.
Of course, you can’t arrest someone for a crime you only believe they might commit. Or at least, not yet. So the Force is using “soft” interventions such as pre-emptive counseling, in a bid to avert future criminal acts.
Insurance
The early naughties were all about gathering big data and the insurance industry was no exception. But while some other sectors, such as retail and banking, have forged ahead with analytics to deliver real business intelligence, insurance has been lagging behind. This is set to change, as AI begins to make inroads.
Until now, a lack of adequate tools has meant that the insurance sector has been challenged not only to handle, store and sort the masses of data generated, but also to understand and place an accurate value on what it has gathered. Decent analytics systems have been lacking.
The potential of AI here is evolutionary rather than disruptive and stands to help the sector catch up with banking in key areas such as risk and intents. If pre-purchase behaviors can be correctly identified through analytics software applications such as those marketed by MRP, lost sales can be avoided and customers met with a response at appropriate places on their journey.
AI can also help with identifying and tackling risk. For example, predictive analytics via machine learning can be put to use when it comes to determining whether an applicant is likely to have certain health conditions, or establishing their true driving proficiency.
Market intelligence
Once you begin to understand the potential of algorithms to transform an industry such as insurance it becomes easy to see how predictive analytics could be used across all sectors.
Market analytics firms — ranging from the conservative, survey-based institutions such as Nielsen and Boston Consulting Group to the more creative “futurologists” — make millions of dollars in consultancy every year by providing hard numbers based on past performance and then extrapolating to make predictions about the future.
They help clients to make decisions on anything from outsourcing a business process to entering an overseas market. But until now, all such consultancies have been able to do is look at past performance and hope it will indicate future performance. They are essentially asking clients to take a leap of faith on their own expertise when it comes to forecasts.
AI predictive analysis stands to turn this model on its head by finally delivering that elusive holy grail: the future in numbers. Time will tell.