Artificial Intelligence for Industries of the Future

Artificial Intelligence has become a ripe topic of discussion, including among business leaders and in the popular press. While some of this discussion has a speculative element to it, much of it is predicated on the enormous successes of AI over the last decade, especially owing to deep learning. This chapter begins with a brief introduction to AI and its relatively young history and the differences between AI, machine learning, and deep learning. We then turn to a discussion of industries of the future and why we prefer that terminology to others. Drivers of industries of the future, including non-AI drivers, are discussed, with real-world examples and citations. We then turn briefly to the interesting question of where AI-based innovations driving industries of the future will likely come from and the important role of fundamental research. In the subsequent chapters, we dive into many of these issues in depth.

Although the academic community has devised rigorous metrics for understanding what makes an AI system better than another in a specific application, it is not always evident that success on such metrics will translate seamlessly to improvement on business metrics, such as increased revenues, cash flow, and profits. As more (expensive) AI projects continue to be proposed, due to both internal and external pressures, business leaders are faced with the need to measure the return on investment (ROI) of such projects or to conduct valuation of such projects rigorously. In part, the problem arises because AI can be challenging to implement properly, and like many emerging technologies, the benefits may not be felt immediately or directly. In this chapter, we discuss the challenges of implementing practical AI systems, many of which stem from data acquisition and quality issues, followed by a deep dive into guidelines and principles for measuring ROI of, and objectively valuing, AI projects.

This chapter provides an extensive dive into AI research and innovation in industry today. We begin by considering the most important source of AI innovation outside of academia: Big Tech research labs. While it is not always evident who should, and should not, be included under the umbrella of “Big Tech,” some candidates are fairly apparent, including Apple, Meta, and Alphabet. These companies and their products have had an outsize impact on smartphones (and other similar devices, as well as enabling services such as digital payment systems and media), social media, and Web search. At the same time, it has become evident over the last several years that companies not traditionally or currently classed as Big Tech companies, such as Tesla, but also large defense companies, such as Lockheed Martin, have also made important innovations in various subfields of AI. We briefly comment on these, before switching our discussion to the Chinese “Big Tech.” Finally, we comment on the important and disruptive role that small- and medium-sized enterprises, including startups, play in fostering and commercializing innovation in emerging technologies such as AI. We close the chapter with a case study on neural language models that have revolutionized applications in natural language processing and potential ethical concerns.

Within the AI research community, there continues to be a focus on greater and more complete automation of difficult problems in application areas such as Computer Vision and Natural Language Processing. However, within industry, there is a greater focus on “augmented” AI. As the name suggests, augmented AI is used to refer to a set of AI-based technologies that are designed to augment, rather than replace, the capabilities of human beings. The difference is more pragmatic than fundamental; in practice, many of the same technologies being developed by researchers are deployed in enterprise as augmented AI. At the same time, we would be hard pressed to ignore the very real ramifications that such technologies have on the workforce and on worker morale. This chapter is an attempt to describe the phenomenon of augmented AI from a business lens. We discuss key features and example applications of augmented AI, using radiology as an illustrative case study. Next, drawing on an influential survey by McKinsey, we discuss trends in organizational behavior and workforce changes that are on the horizon. Of particular interest is a new way of working, and a new type of job called a “new collar” job. The chapter also discusses adaptation that will be required in the C-suite as a consequence of these changes and draws on examples from three industrial sectors.

With rapid growth and adoption of emerging technologies like AI, ethical use of such technologies become paramount. Many such ethical principles also get formally codified into law and policy. We begin this chapter by differentiating between AI and digital ethics, the key difference being that the latter tends to have broader scope than the former. We then dive into the philosophy of ethics, followed by a discussion of how AI ethics is being incorporated into policy. Several countries are used as real-world examples, but to illustrate such policies in depth, we provide two case studies. The first of these is on the influential and much-discussed General Data Protection Regulation (GDPR) enacted in the European Union in the last decade. Although it is still controversial, and perhaps too early to say, whether enforcement of GDPR has been sufficiently strong or effective, the regulation has already been used to administer a number of fines and penalties on large corporations like British Airways and Marriott. The second case study is on the US-based National Defense Authorization Act (NDAA). We close the chapter with a discussion of AI ethics in higher education and research.

This chapter concludes the book by covering a set of issues that we believe will prove to be important for any industry or organization that is actively seeking to incorporate AI systems into their pipelines. We begin with the seemingly outlandish issue of AI copyright that has rapidly become important due to state-of-the-art AI’s success in generating “creative” art and writing. The central question is whether an AI should be allowed to own its own copyright. Related to the question of copyright is one of regulation. Currently, there is a vigorous debate on both regulation of algorithms, as well as regulation by algorithms, since the latter may yield more objective outcomes. Critics point, however, to the dehumanizing effects of such regulation, and to the fact that such algorithms may not be objective to begin with, since they have been trained on (potentially biased) human-generated data. We also cover the important and mainstream issue of legal regulation of Deepfakes and AI’s “explainability crisis” and conclude the chapter and the book with a note on the rapid convergence of AI with other emerging technologies, such as quantum computing.