AI: Not artificial anymore. Incorporating artificial intelligence (AI) and machine learning (ML) into business processes creates an intriguing prospect. With finance being one of the most critical functions of an enterprise, CFOs should understand and leverage AI and ML to provide real time insights, inform decision making and drive efficiency across the enterprise
There is a subtle difference between AI and Machine Learning. AI is a branch of computer science attempting to build machines capable of intelligent behavior, while machine learning can be defined as the science of getting computers to act without being explicitly programmed. In another words, AI researchers build the smart machines, while machine learning experts would make them truly intelligent.
Deep learning, a further subset of machine learning gaining lot of prominence of late, imitates the workings of the human brain in processing data and creating patterns for use in decision making. Facebook’s use of face and image recognition is an example of Deep learning.
AI and machine learning are already driving the technology we use in our everyday lives. For example, typing the first few letters of a query into Google and having the remainder anticipated is a result of machine learning, as are recommendations from Netflix on what to watch next. Similarly, driverless cars, smart personal assistants such as Siri, Cortana, and Alexa are some of the common examples of AI applications. Incorporating AI and machine learning into business processes creates an intriguing prospect.
AI and ML: Promising applications in Finance
Managing Portfolio
Algorithm-based robot-advisors are built to calibrate a financial portfolio to the goals and risk tolerance of the user. Users fill in their goals (for example, retiring at age 62 with $350,000.00 in savings), age, income, and current financial assets. The robot-advisor then calibrates to changes in the user’s goals and real-time changes in the market. These AI backed robot-advisors have gained significant traction with millennial consumers who don’t need a physical advisor to feel comfortable investing.
Algorithmic Trading
Algorithmic trading utilizes advanced and complex mathematical models to make high-speed transactions in and determine trading strategies for optimal returns. Most hedge funds and financial institutions do not openly disclose their AI approaches to trading, but it is believed that machine learning and deep learning plays an increasingly important role in calibrating trading decisions in real time.
Fraud Detection
While earlier or conventional financial fraud detection systems relied heavily on complex and exhaustive sets of rules, modern fraud detection goes beyond following a checklist of risk factors – it continuously and actively learns and calibrates to new security threats. By using machine learning for fraud detection, systems can detect unique activities or behaviors and flag them for security teams
Loan / Insurance Underwriting
Machine learning algorithms can effectively process and get trained on millions of examples of consumer data such as age, job and marital status, as well as financial lending or insurance results including whether an individual defaulted or paid back a loan on time. Algorithms can continuously analyze and sense trends that might influence lending and insuring in the future. Good example is the application from RNDpoint for bank credit systems and Fintechs. Contact Viktor Prilutskiy
Customer Service
Chat bots and similar conversational solutions are a rapidly expanding area of investment in customer service budget.
These virtual assistants are built with robust natural language processing engines as well as the nuances of finance-specific customer interactions. Banks and financial institutions that provides such a swift querying and interactive experience might pick up customers from traditional banks that require people to log into a time -consuming online banking portal and do the digging themselves.
Sentiment Analysis
Much of the future applications of AI and machine learning will be in understanding social media, news trends, and other data sources – not just stock prices and trades. The stock market moves in response to numerous human-related factors, and the ability of AI and machine learning to process and understand their large data sets will one day be able to replicate and enhance human financial intuition by discovering new trends and telling signals.
New Security Norms
The current personal security features like login credentials and security questions may no longer be the norm for user security in the coming years. In addition to glitch-detection applications like those currently being developed and used in fraud, future security measures might require facial recognition, voice recognition, or other biometric data, powered by AL and ML in the background.
Intelligent Approval workflows
Currently, approval workflows mostly include matrices that list various conditions based on which approval levels are triggered. But these approval workflows don’t consider the broader circumstances, like if the requester is new in role and might require more supervision, or whether previous request from this requester been rejected or approved. AI-based intelligent workflows could allow finance team to distinguish and filter out the true exceptions from the standard low-risk exceptions that are usually approved anyway. This way, employees do not need to wait for approvals and feel empowered, while still limiting the risk to the corporation.
AI and Machine Learning: The way forward for the CFO
Surprisingly, AI and machine learning are still not on the radar for many CFOs as part of their strategic future investment areas.
This might limit their long-term chances of either maintaining or achieving strategic positioning in the market or among their competitors. It’s similar to when cloud technologies emerged as a potential disruptor to the accounting profession. Many didn’t foresee that cloud adoption would become so widespread, but it has now become standard. AI and machine learning present an even larger potential disruption.
A lot of accounting technology companies are experimenting with AI-based solutions and implementing them in their platforms, something forward-looking CFOs should be optimistically in. Some next-generation applications powered by machine learning can significantly optimize the cash application process by continuously analyzing historic data such as pay patterns, behavior and clearing documents, and based on this information update matching principles to clear payments automatically.
With this approach, the efficiency and effectiveness of cash application can be improved significantly and achieve extraordinary automation rates.
The AI and machine learning business opportunities have only just begun to scratch the surface of what’s possible. Many companies have already initiated their strategic investments in this field. CFOs should begin considering their company’s investment strategy on these future technologies.
The smart way for handling the risks of AI and machine learning
While there are immense possible business applications for these two technological trends, CFOs should also be aware of the associated risks.
A simple example of unintended consequences is price discrimination. A machine, still not evolved that much, cannot make moral judgments about discrimination; it can only make decisions about classes of customers with no understanding of who is part of a marginalized group or what the legal implications might be. As more decisions become automated, the risk of having conflicts with laws and regulations increases if these applications are not fine-tuned during implementation stage. One wrong decision by a machine might not amount to much, but they can have a material financial impact in aggregate. It’s the CFO who will have to answer for any fiscal impact, making it important for them to understand how the algorithms operate, make decisions, and what the ramifications might be for shareholders.
In spite of the initial risks associated with any new technology, AI and Machine learning is shaping up to be the next major evolution in the transformation of finance CFOs should prepare for. CFOs might want to explore the following ways to unlock the value machine learning has to offer.
Start experimenting with data
Machine learning is about data experimentation, hypothesis testing, fine tuning data models and automation. CFOs should consider using innovation labs, ideation forums, and create skunk work project teams where developers can bring together a discrete data set that hasn’t been tested before and use machine learning to identify hidden patterns. It will help assess the potential risks, before they are put into production.
The National Health Service in the United Kingdom delivers healthcare to more than 60 million citizens of the UK. By using AI to learn more from its huge volume of patient data, they redesigned the health card application process over three months by using variance detection to find fraudulent activity. By delivering value in a short timeframe, they received backing to expand. Now they have a long-term strategic goal of saving £1 billion over 5 years.
Put data under the business ethics lens
At the 2018 Gartner Business Intelligence & Analytics Summit in Munich, Gartner shared an estimate that half of business ethics violations will occur through the improper use of big data analytics by 2020. This can lead to a loss of reputation, limit business operations, losing out to competitors, inefficient or wasted use of resources, and legal sanctions.
Therefore, CFOs should examine all the potential ramifications before putting their experimented data findings into practice, including any legal, financial, and brand implications. Ideally, an expert committee on business data ethics should audit algorithms for unintended consequences, thus reducing the risks associated with machine learning.
Figure out the high-value data
Today, due to the humongous rate of data generation, even small to midsize companies collect far more information than they can ever utilize. Only a fraction of it will ever hold predictive value. So, CFOs should carefully decide which data might be worth something, and which data sets can be discarded. Some of it may need to be kept for regulatory purposes; others, for commercially useful predictions and products. Keep only what is needed and what is potentially valuable.
Identify processes where AI and ML can bring value: better, faster and cheaper
CFOs must continually make choices about how they allocate resources. There’s always internal tussle for funding to pursue new business opportunities, and an investment in one area requires savings in another. AI-based machine learning can enable increased savings by taking automation to a much higher level than previously possible.
In many companies, a high percentage of staff still perform transactional tasks that can be automated through machine learning. By letting self-learning algorithms find patterns and solutions in data instead of following preprogrammed rules, transactional tasks can be completed exponentially faster and with fewer people.
Back-office processes like procure-to-pay, order-to-cash, and record-to-report can be radically automated as business networks eliminate manual work.
“In many companies, a high percentage of staff still perform transactional tasks that can be automated through machine learning.”
Invest in developing future skills set
The real value in AI and machine learning are about gaining control, identify pain areas and bringing improvement by using advanced AI-based business solution, to drive the business forward, which is one of the fundamental responsibilities of CFOs.
For CFOs, it also means automating as much as they can, moving away from the traditional accounting tasks of performing transactions, reconciling accounts, and compiling reports. With the automation of transactional tasks, CFOs and their teams can focus on partnering with the business to analyze available data, identify new business opportunities, and provide strategic guidance. At the same time, they must consider how to train, develop, and create a future-ready talent pool for changing business models. By collaborating with professional accounting bodies, CFOs can offer continuous learning opportunities to their critical talent pools.
Align finance to the overall digital strategy of the enterprise
CFOs need to actively start taking part in the organization’s discussions about digital transformation. Being part of this strategic conversation helps generate the required momentum and reduce resistance to change during the implementation phase.
CFOs must be adequately aware of the expected digital angles to help solidify the organization’s digital strategy so that when a business case is up for review, they are well informed and can make the right decisions.
As AI and machine learning evolves, CFOs should make proactive efforts to familiarize themselves with its business opportunities.
AI and ML: Implications for the present
AI and machine learning are no longer the upcoming trend of the future. With significant investment and technology maturity advancement already started, many firms are actively exploring the technology and identifying practical ways it can impact the business. Some have already begun adoption, as revealed by the prediction that the market for AI-based solutions will experience a compound annual growth rate of 55.1% over the 2016-2020 forecast period. Furthermore, almost 25% of today’s jobs are expected to be impacted by AI technologies by as soon as 2021.
For CFOs, that means they need to observe these vital indicators not just for future investment decision making but also drive discussion to bring them into the digital transformation objectives.
A true digital transformation program requires more than just applying the latest technology. It needs a customer-focused, outside-in perspective to empower the design of digital solutions that can drive customer loyalty, engagement, consumption and satisfaction. AI and machine learning can be the key to providing the capability, insight and acceleration that enable tomorrow’s business to thrive in this environment
Part two
The proliferation of robots completing manual tasks traditionally done by humans suggests we have entered the machine automation age. And while nothing captures the imagination like self-directing machines shuttling merchandise around warehouses, most automation today comes courtesy of software bots that perform clerical tasks such as data entry.
Here’s the good news: Far from a frontal assault on cubicle inhabitants, these software agents may eventually net more jobs than they consume, as they pave the way for companies to create new knowledge domain and customer-facing positons for employees, analysts say.
The approach, known as robotic process automation (RPA), as we discussed in my former blog about RPA automates tasks that office workers would normally conduct with the assistance of a computer.
Bots mimic activities a human would perform, including anything from populating electronic forms to changing data in a customer account. Some bots log into an application, extract information from a web page, modify it and enter it into another application. At AT&T, bots pull sales lead from multiple systems, enabling staff to spend more time with customers.
Rise of the machines yield greater productivity
Bot benefits include the capability to cut staffing costs, reduce error rates associated with humans and improve customer engagement. For example, a bank redesigned its claims process and deployed 85 bots running 13 processes, handling 1.5 million requests per year. The bank added capacity equivalent to more than 200 full-time employees at approximately 30 percent of the cost of recruiting more staff.
Man vs. Bot
A major appeal of bots is that they are typically low-cost and easy to implement, requiring no custom software or deep systems integration. Such characteristics are crucial as organizations pursue growth without adding significant expenditures or friction among workers. “Companies are trying to get some breathing room so they can serve their business better by automating the low-value tasks.
There’s little question that many workers will lose their jobs as companies automate more business processes. Forrester Research last November estimated that RPA software will threaten the livelihood of 230 million or more knowledge workers, or approximately 9 percent of the global workforce. However, RPA will create new jobs as workers train up and pivot to new roles within their companies.
RPA, along with physical, intelligent machines and other automation capabilities, will replace 16 percent of jobs but create the equivalent of 9 percent, yielding a net loss of 7 percent of jobs by 2025, a Forrester analyst who tracks the impact of automation technologies on the corporate sector. For instance, Wolters Kluwer reallocated money saved using RPA to close its books to hire a financial analysts to analyze profits, revenue, planning and forecasting.
Increased RPA will give rise to “cognitive sommeliers,” or staff who understand domains and curate knowledge bases for an application area. Moreover, as the glut of information increases, particularly in financial services, customers will need more human advice than ever before.
Most bots stick strictly to their business logic rules but that is changing. If the machines can become smarter, the popular thinking goes, businesses will be able to use them in more complex operations. Paired with chatbots, natural language processing, machine learning and other tools, RPA can extract and structure information from audio, text, or images, as well as identify patterns and pass that information to the next step of the process.
Bots are already making a difference in how businesses interact with customer. At Vanguard Group, sophisticated algorithms called “roboadvisors” pair with humans to offer clients tailored investment advice. Virgin Trains has deployed cognitive RPA to automatically refund customers for late running trains. As customer emails arrive, a natural language processing tool gauges meaning and sentiment and then recognizes key information in the text to service the customer, reducing daily processing time and manual labor involved in dealing with customer emails by 85 percent. Cognitive RPA has the potential to go beyond basic automation to deliver business outcomes such as greater customer satisfaction, lower churn, and increased revenues.
Change management and integration challenges loom
However, cautions, cognitive RPA has been slow-going due to the complex nature of blending the technologies with existing systems, as well as the lack of required skills to implement them. To bridge that talent gap, IBM and Blue Prism have inked a joint agreement to work together on cognitive RPA.
Organizations piloting the most basic bots must clear change management, governance and security hurdles. Implementing too many interdependent bots can wreak havoc on existing systems. Bots can also create turmoil when paired with workers who must learn to work with their new virtual colleagues.
Moreover, because most RPA tools reside on desktops, implementations in environments that are highly virtualized — where information from thousands of PCs resides on centralized servers — can be clumsy. They have integration problems with more sophisticated VDI [virtual desktop infrastructure] implementations.
CIOs should introduce RPA quickly in increments but scale it up slowly. Start small and start fast. “If you see early success, you need to take a step back and start thinking more strategically about how you scale up in terms of governance and your staffing model.”
Personal advice to all involved with Digital Transformation:
And in all cases remember this: small teams, transparent communication abroad the entire organization will bring less obstacles and keeping all participants in line. This happens with the latest failure at Nokia. Whatever happens learn from this and communicate from top-down level, because when a CEO, CFO communicates this from top level the rest will follow.
