Machine Learning - eBook (EN)

E X E C U T I V E G U I D E THE AI & MACHINE LEARNING IMPERATIVE MIT SLOAN MANAGEMENT REVIEW 2 Tax, audit, and compliance functions are also well suited for AI. AI can extract supplier contract terms and match them to goods and services deliv- ered. Employee expense report information can be checked against corporate policies. The speed and constant work of AI systems typically means that 100% of transactions can be audited rather than just a sample. Other forms of AI can be used for forecast- ing, estimating demand (typically using external data), and assessing risks, including conducting appraisals of brand damage from problems such as security breaches. The use of AI for finance functions is taking root. In the 2020 global AI adopter survey, 8% of respondents put finance in the top two application domains. But in a 2018 Deloitte survey of enterprise AI, 37% of large U.S. organizations were pursuing use cases with AI involving risk management, 29% on cases involving forecasting applications, and 23% on cases addressing tax, audit, and compliance is- sues — all of which touch on finance. Admittedly, AI applications in finance thus far have been relatively pedestrian. However, the future of intelligent financial applications is likely to be much more dramatic. Most transactions will be automated, replacing outsourcing as a way to achieve productivity. Finance functions will likely be staffed by substantially fewer people, all of whom understand AI and how to add value to it. Budgets, forecasts, financial analysis, and approaches to improving financial performance will likely be based on machine learning models trained by internal and external data. These developments will be particularly im- portant in the COVID-19 economy ; some AI companies, for example, have already developed ap- proaches to using AI to minimize accounts payable cash outflows. Some companies are basing rapidly changing demand predictions on external data such as surveys, mobile phone data on consumer move- ment away from home, and even sensor data from consumer thermometers. Many of the external services offered to finance functions — auditing, consulting, and tax advice — will also be automated and substantially more intelligent; Deloitte, for example, is rolling out AI- supported audits now. In 10 to 20 years, we expect that CFOs will oversee a stable of algorithms and AI applications that will make their function more suc- cessful and efficient than it has ever been. Opportunities for AI Within Finance-Related Functions CFOs who lead business functions such as IT and procurement may be responsible for sponsoring and overseeing AI applications in these areas. AI, of course, can ultimately can lead to substantially greater productivity in IT functions, especially in industries where IT is becoming the "factory" for financial services and e-commerce businesses and others. However, Deloitte's 2020 enterprise AI sur- vey found that 62% of executives view cybersecurity as one of their top three concerns about AI, so CFOs may want to play a leadership role there. A variety of AI applications can be employed in the procurement domain, including spend clas- sifications, supplier risk assessments, automated contract reviews, and chatbots for routine supply ordering. Even when procurement does not report to the CFO, it may be wise for finance chiefs to pay attention to this area and insist on some of these capabilities because of the implications they could have on the financial health of the business. Oversight of AI Investments CFOs usually have the oversight role for their com- panies' spending and investing activities. Many businesses are spending significant amounts on AI — 53% of respondents to the Deloitte 2020 survey said their companies had spent more than $20 million an- nually on AI technology and talent, and 71% report that they will spend more on AI in the coming year. The Deloitte survey data also suggests a large majority — 81% — of the longer-term users of AI have seen returns on their AI investments, with a payback period of less than two years. CFOs can help ensure that high levels of value from AI continue to be achieved by creating sys- tems and processes for reviewing investment proposals, moving AI systems toward production deployment, and assessing the value of systems post-implementation. These moves are particu- larly important in the difficult economic climate we are likely entering. Despite the fact that AI is a relatively new technology, each investment should be intended to deliver financial value to the organization — even though not all of them will pay off. CFOs can play a role similar to that of a venture capital partner, doing whatever is necessary to clear the way for AI initia-

• Cover