Artificial Intelligence (AI) is simultaneously an older, established technology and a newer, rapidly developing and emerging technology. How can it be both old and new? The ability to write software programmes that could do more than “compute” numbers — that could use AI methods for logical inference to prove logic theorems and solve algebra problems with symbols instead of numbers, goes back to the 1950s. The first chatbot was created in the mid-1960s using the AI methods available at that time. Knowledge-based systems using “If-Then” rules proliferated in the 1980s and were used to address and support a wide variety of practical industry problems. Many rule-based systems are still in operation today. All of these earlier applications of AI were based on the ability to manipulate abstract symbols and codify and manipulate knowledge.

Since about 2012, a new generation of AI methods with new types of capabilities have swept across the world and have been used in every industry and type of application imaginable, and then some. In fact, this newer and more recent wave of AI applications trace back to principles and methods that have been around for decades. However, the combination of such older methods with the explosion of data in digital form, dramatic increases in hardware capabilities, reductions in the cost of computation, and the rise of new (from the 1980’s, 1990’s and onward) mathematical algorithms for processing this data led to the common impression that AI is only recently emerging as a “new technology”. That’s not true. What is true is that the capabilities and scope of AI have dramatically improved in recent years. And, by the way, those older AI methods from the 1950s through 1980s have not disappeared. They are still in use and play an important role, though now mostly in combination with the newer, data-driven machine learning methods which have become so central to modern AI.

All of these new developments with AI methods and real-world applications are indeed  changing how we work, live and play. Any person using a smart phone or shopping on a major e-commerce site is benefiting from AI applications as part of their daily lives. Many new applications are being piloted or are already underway, such as new types of virtual assistants or personal healthcare advisors which are more capable but still limited in capability.  There are companies in every industry worldwide using AI methods to improve their products, services and internal operations, and the number of companies making use of AI is rapidly growing. At the same time, only a relatively small portion of all companies that are potential users have adopted AI methods into their everyday processes or their products Therefore, we are still early in the diffusion cycle.

There is no end in sight to this expansion of usage of AI methods across the economy of every country as there are so many different types of AI methods, so many ways in which these methods can be combined, and so many demonstrated as well as potential ways to make use of these methods to improve business performance, services and products.

In this context, Tom Davenport, President’s Distinguished Professor of IT and Management of Babson College and Digital Fellow at the Massachusetts Institute of Technology Initiative on the Digital Economy, together with Steve Miller, Professor Emeritus of Information Systems and former Vice-Provost (Research) at Singapore Management University, set out to understand how people are actually doing their everyday work in collaboration with AI systems in the context of real-world work settings across knowledge and service work, factory work, and field work. Their full set of 29 case studies and insights derived from these examples will be published in the forthcoming book, “Working with AI: Real Examples of Human Machine Collaboration” by MIT Press in the second half of 2022.  While they were in the process of preparing their material for this book, they shared some of their interim findings based on 24 case studies in a summary article they published in May 2021 in SMU’s Asian Management Insights magazine. The remainder of this write up is based on that summary article.

Through their case studies, the authors observed and described how the integration of AI into work processes impacts the changing nature of work. They provided examples of how AI system usage in actual operational settings is being tapped to reduce costs and provide greater worker productivity. In the case examples they studied, the employers did not reduce employee headcount as a result of the productivity improvements enabled by using the new AI capabilities. Rather, the employers boosted productivity by expanding the amount of work they could do or the range of work they could do with their existing staff. Their findings are especially useful for companies considering the adoption of AI, or while planning or deploying these practices.

Citing a  study by the consulting firm Kearney that was commissioned by EDB International, Davenport and Miller noted that the current AI adoption rate is slow in Southeast Asia, with more than 80 per cent of companies still at the early stages of using this new technology. However, the region is on the cusp of a major technological revolution as AI applications in business are beginning to emerge: the top five economic sectors–manufacturing, retail and hospitality, agriculture, healthcare, and government (including safety, security, and smart cities)–will benefit from increasing AI usage to drive strong overall impact. (See  AI adoption in Southeast Asia.)

Here are some of the key insights from Prof Davenport’s and Prof Miller’s May 2021 summary article  highlighting how working with smart machines can elevate the future of work:

1. Determine the right platforms

Platforms are the support systems that acquire, integrate and manage data employed in AI applications, and ensure the effective operation of an AI system. The choice of platform is critical in enabling a smooth flow, from providing the inputs for the AI application to handling the outputs. Moreover, they have strong implications on job roles within the organisation: For example, personnel in both IT and business roles involved in creating and maintaining these platforms need to collaborate with the internal or external data scientists who design and implement the AI algorithms at the foundation of these platforms.

Stakeholders should consider the best platform required for individual AI applications, and if the company ought to build a number of single-use platforms or fewer multi-functional ones. By examining the feasibility of acquiring or embedding AI capabilities into existing transaction platforms, as well as the type and degree of human involvement involved, they can also better formulate an AI strategy to fit their goals. Finally, organisations should also examine how the choice of platform impacts workforce skills and change job roles, and if they have the right roadmap for optimising AI technology with relevant human capital.

2. Provide adequate training of human users

When employing previous generations of case management systems (CMS), workers had to assimilate information, assess the situation and make their judgments and decisions. Such systems provided information support rather than delivered system-driven decision recommendations.

Today, AI-enabled CMS can use available data and automated decision-making algorithms to make recommendations or even preliminary decisions. While such intelligent systems can automate workflows and increase productivity in organisations, the human user needs to be equipped with a practical understanding of the system’s decision-making process and the data it relies upon to do so, before they can accurately review, modify, or override the decisions by the smart system.

3. Embrace human-machine collaboration

An advantage of humans and smart machines working hand-in-hand is that humans can affirm that an automated decision is appropriate for the specific context and circumstances at hand.

A tool called ShotSpotter Connect, for example, can recommend areas in which a police officer should patrol, and what to do while patrolling based on data analysis by a smart machine. The ultimate decision-making power, however, lies with the user — but they will make an informed choice using a suggested action plan as guidance.

In many work settings, the ‘right’ decision often depends on understanding how contextual factors and contingencies can change a situation significantly, which cannot be deduced solely from data-driven decision-making. As such, it can be risky to rely completely on intelligent computer-based systems to automate decision-making without human review, and the study authors believe the final output is usually better with the combination of human and machine expertise.

4. Work from anywhere

Automation is one of the top benefits of adopting AI systems, such as the use of intelligent computer-based CMS to eliminate manual paper flows, or integrate inputs from disparate online data sources and support tools, and streamline workflow.

Thanks to online CMS, users like frontline workers and their supervisors can work anywhere and anytime. All the resources required for getting work done are present in the online applications. This proved to be especially useful during the workplace restrictions resulting from Covid-19 countermeasures.

However, while workers enjoy greater autonomy, remote work also gives rise to other considerations. For example, employees will require access to relevant technologies in their homes. Their work is also mediated and monitored by software, which might lead to some personnel feeling “chained to the computer”, or complaining that “the work never stops”. The authors  therefore, advise employers to mix this form of work with other tasks that could integrate social activities and non-computer work, to make jobs more fulfilling and avoid employee burn-out.

5. Explore new job specialisations

The increased adoption of AI has also led to the growth of new job specialisations. The authors, for example, has noted a rise in the “hybridisation of business-related roles with IT and AI deep tech-related roles”. This includes employees with business backgrounds in IT and related tech roles (including the Chief Information Officer role), as well as people with deep tech (including AI and analytics) backgrounds being embedded into business units and other non-IT corporate groups.

And this need for companies to build deep technical expertise in IT, AI, cybersecurity, and data protection continues to increase. Concurrently, there is a need for even more human capital to step into these hybrid business-tech fusion roles. Indeed, new AI developments are proceeding at breakneck speed. But bringing everything together across technology, people, and job roles in any real-world work setting is a very complex undertaking.

Companies committed to harnessing AI-enabled strategies should begin by identifying and engaging the complex ecosystem of stakeholders, participants and other companies who will be impacted. Thereafter, business leaders can zone in on the potential of AI to enable their organisations to thrive in an increasingly competitive marketplace, and help enhance operations quickly without many costly errors or delays, while also reducing costs significantly over its lifespan.

This summary is based on the article by DAVENPORT, Tom and MILLER, Steven “Working with smart machines: Insights on the future of work” (2021). Asian Management Insights. 8, (1), 18-25. Research Collection School Of Computing and Information Systems.
Available at: https://ink.library.smu.edu.sg/sis_research/5930 

We thank Professor Emeritus Steven Miller for his help on this write up.