AI for Real-World Applications

Published on 20 May 2022

AI for Real-World

Artificial Intelligence for Real-World Applications

In 2013, the MD Anderson Cancer Center initiated a "moon shot" effort to use IBM's Watson cognitive technology to diagnose and prescribe treatment regimens for some kinds of cancer. In 2017, however, the project was halted since its expenditures had exceeded $62 million, and the technology had not yet been utilized on patients. Simultaneously, the cancer center's IT department was experimenting with cognitive technologies for considerably less ambitious tasks, such as recommending hotels and restaurants to patients' families, assessing whether patients required assistance paying their bills, and resolving staff IT issues. These initiatives have produced significantly more promising results: The new technologies have led to higher patient happiness, enhanced financial performance, and a reduction in the amount of time care managers spend on monotonous data input. Despite the failure of the lunar mission, MD Anderson is dedicated to employing cognitive technology — that is, the artificial intelligence of the next generation — to improve cancer therapy, and is now developing a number of new initiatives at its center of competence for cognitive computing.

The distinction between the two techniques is crucial for anybody developing AI projects. Three-quarters of the 250 executives acquainted with their firms' usage of cognitive technology who participated in our poll think that AI will significantly impact their organizations within three years. However, our examination of 152 initiatives in nearly as many organizations suggests that "low-hanging fruit" projects that improve business processes are more likely to succeed than extremely ambitious "moonshot" efforts. This should not come as a surprise, since this has been the situation with the vast majority of new technology embraced by businesses in the past. But the hype around artificial intelligence has been exceptionally potent, and it has captured certain businesses.

This post will examine the many kinds of AI now in use and present a framework for how businesses might begin to grow their cognitive skills in the next years to meet their business goals.

Also Read: U.S. Court Denies Facebook's Motion To Dismiss Antitrust Case

Three Forms of AI

It is advantageous for businesses to view AI through the prism of business capabilities as opposed to technology. AI can assist with three essential business requirements: automating corporate operations, getting insight via data analysis, and engaging customers and workers.

Automating processes

The majority of the 152 projects we evaluated included the automation of digital and physical processes, primarily administrative and financial back-office chores, employing robotic process automation technology. RPA is more complex than previous business-process automation solutions because the "robots" (that is, code on a server) mimic human behavior by entering and consuming data from numerous IT systems. Activities include:

moving data from e-mail and contact center systems into systems of record, such as updating client files with new addresses or new services;

replacing lost credit or ATM cards, accessing numerous systems to update records, and managing customer contacts; resolving failures to charge for services across billing systems by extracting data from different document kinds; and

Legal and contractual papers are "read" using natural language processing to extract provisions.

RPA is the least costly and simplest to adopt of the cognitive technologies we'll explore here, and it often yields a high rate of return. (It is also the least "intelligent" in the sense that these apps are not built to learn and improve, but developers are gradual including more intelligence and learning capabilities.) It is well suited for integration with numerous back-end systems.

NASA launched four RPA pilots in accounts payable and receivable, IT expenses, and human resources, all administered by a shared services center, in response to cost challenges. The four successful projects—86% of transactions in the HR application, for example, were performed without human intervention—are being implemented throughout the firm. NASA is now deploying additional RPA robots, including those with more intelligence. According to Jim Walker, the project manager for the shared services group, "it's hardly rocket science so far."

One could assume that robotic process automation will eliminate jobs rapidly. In contrast, among the 71 RPA initiatives we examined (47 percent of the total), the replacement of administrative staff was neither the main aim nor a frequent effect. Few initiatives resulted in reductions in headcount, and in most situations, the relevant duties had previously been outsourced. Future robotic automation initiatives are anticipated to result in some job losses, notably in the offshore business-process outsourcing sector as technology advances. If work can be outsourced, it can likely be automated.

Cognitive insight

The second most prevalent kind of project in our survey (38 percent of the total) used algorithms to find ways in massive amounts of data and analyze their significance. Consider it to be "analytics on steroids." These applications for machine learning are used to:

  • determine what a certain buyer is likely to purchase; identify credit card and insurance claims fraud in real-time;
  • Analyze warranty data to discover vehicle and other manufactured goods safety or quality issues; automate tailored digital ad targeting; and
  • Provide insurers with more precise and comprehensive actuarial modeling.

In three ways, the cognitive insights supplied by machine learning vary from those provided by conventional analytics. They are normally far more data-intensive and detailed, the models are typically trained on a portion of the data set, and the models improve over time; that is, their ability to utilize fresh data to generate predictions or classify things increases.

Cognitive Involvement

In our survey, projects that engage workers and consumers via the use of chatbots, intelligent agents, plus machine learning were the rarest (accounting for 16 percent of the total). This category contains:

  • Intelligent agents that provide customer assistance 24 hours a day, seven days a week, handling a wide range of concerns, from password queries to technical support inquiries, in the client's native language;
  • Sites for addressing employee inquiries about IT, employee perks, and HR policies;
  • Service and product recommendation systems that boost personalization, engagement, and sales for merchants, generally via the use of rich language or graphics; and
  • Health treatment suggestion systems that aid in the creation of individualized care plans for patients based on their health state and past treatments.

See also: Meta Was Anointed 'Worst Company' of 2021 In The US

Companies in our research tended to communicate with workers using cognitive engagement technology more often than with consumers. This may alter if corporations get increasingly comfortable with automating consumer interactions. For instance, Vanguard is launching an intelligent agent that assists its customer support workers in responding to commonly requested inquiries. Eventually, consumers will be able to interact directly with the cognitive agent, rather than with human customer care employees. SEBank in Sweden and the medical technology business Becton, Dickinson in the United States use the realistic intelligent-agent avatar Amelia to provide IT assistance to their internal employees. SEBank has just made Amelia accessible to a limited number of clients in order to evaluate its performance and reception.


Featured image: Brain technology photo created by


Subscribe to to learn about new updates and changes made by tech giants that affect health, marketing, business, and other fields. Also, if you like our content, please share on social media platforms like Facebook, WhatsApp, Twitter, and more.