Application Programming Interfaces (API) and AI

An application programming interface (API) is a set of functions or rules that facilitates communication between applications, databases, and devices. An analogy to understand how an API works is a restaurant waiter who takes an order relays it to the kitchen and then returns the prepared food to the customer. APIs are like digital messengers - they allow the weather app on your phone to transmit and retrieve information from a third-party app such as the Weather Channel. As another example, APIs enable customers to purchase movie tickets through a site like Fandango by signing into their Facebook or Google accounts.

APIs are used extensively by Al and ML programmers. For example, Google’s Prediction API provides cloud-based access to machine learning, natural language processing, recommendation engines, pattern recognition, and prediction services (Rizk, 2018). In a blog post entitled “Top 20 APIs You Should Know in Al and Machine Learning,” Oleksii Kharkovyna (2019) notes that his preferred APIs are most suitable for Al and ML purposes. BigML, for example, uses APIs to create data sets, models, and predictions that use ML technology. As another example, Anaconda Enterprise is a data science platform that uses APIs to deploy machine learning models.

Best Programming Languages Accountants Should Learn for Artificial Intelligence Applications

Accountants with a strong technical background in programming languages will be best positioned to work with data scientists in implementing Al initiatives. For those do-it-yourself accountants who may want to customize Al technologies to solve applications, learning a programming language will be quite useful. Proficiency with popular open-source programming tools for Al, such as Python and R. will enable accountants to customize Al applications to suit the unique business needs of their firms or clients. Other tools, such as SQL and

NoSQL, will be useful for retrieving, editing, and manipulating data to be used with Al applications.

Accountants without a strong background in programming languages can also help develop АІ-enabled applications by using “nocode” and “low-code” platforms. These platforms use drag-and-drop editors for application development rather than traditional coding languages. Some of the world’s leading technology companies offer these solutions. For example, the “Create ML” application by Apple allows users to develop and train ML models using different datasets. Google’s Al Platform supports “Kubeflow”, their open-source application that enables developers to build Al/ML projects. Lastly, Microsoft’s Cognitive Toolkit “CNTK” enables applications built for the Azure platform and allows users to develop customized applications for computer vision, natural language understanding, facial recognition, and prediction models. No-code and low-code tools are undoubtedly useful for accountants in implementing Al initiatives.

Notes

• 1 As used throughout this book, "KMPG" refers to KPMG International Cooperative (“KPMG International”). Member firms of the KPMG network of independent firms are affiliated with KPMG International. KPMG International provides no client services.
• 2 As used throughout this book, "Deloitte” means Deloitte & Touche LLP, a subsidiary of Deloitte LLP. Please see www.deloitte.com/us/about for a detailed description of the legal structure. Certain services may not be available to attest clients under the rules and regulations of public accounting.
• 3 As used throughout this book, "KMPG" refers to KPMG International Cooperative (“KPMG International”). Member firms of the KPMG network of independent firms are affiliated with KPMG International. KPMG International provides no client services.