In October 2018, a Reuters article informed the world that Amazon had scrapped an AI–based recruitment application that turned out to be biased against women. Most headlines about this story highlighted the company’s failure in developing an actionable and fair solution for one of the most important processes of the HR team.
However, what this and similar examples of today’s AI “failures” neglect to acknowledge is the complexity of end-to-end process automation based on AI technology. This complexity stems not only from current technical limitations but also from the immaturity of corporate policies, government regulations, and legal systems to deal with machines that automatically analyze, decide, and act.
This is not to say that AI researchers and some commercial vendors haven’t produced impressive results with various machine learning, deep learning, and reinforcement learning techniques applied to vast amounts of image, video, audio, text, language, and operational data in both enterprise and consumer settings. However, most of the successes have been based on using AI to automate specific low-level tasks.
Learning from Amazon’s AI Automation Experience
In fact, the Amazon recruitment application example shows that its AI developers were quite successful in creating an application that could read massive amounts of text, interpret it, and deliver recommendations to human recruiters. When we view these as discrete tasks, it’s reasonable to argue that Amazon succeeded in automating them using AI. Where AI was accused of failure was in automating the whole process that strings together tasks and activities and constrains and governs them according to today’s regulations and societal norms.
An alternative headline to this story could’ve been “Amazon succeeds in applying human governance to AI-based automation evolution.” Amazon representatives said that this AI-based application was never put into production. A process was in place that allowed people, who saw the machine’s recommendations, not to act on them.
There are two lessons in this example. First, is to understand the scope of AI-based automation. A machine can be great at applying AI to automate a task yet fail at automating whole processes, Second, is that the automation of processes will not be instant but rather evolve through levels of a changing relationship between the human and the machine. It is this understanding that can help decision makers cut through the noise of today’s AI headlines.
Introducing IDC’s AI Automation Evolution Framework
To help with planning and investment decisions related to AI-based automation, IDC has developed a framework that allows for the evaluation of the human-machine relationship and its application at different levels of automation scope across any industry and functional use case. IDC’s AI-based automation evolution framework has five levels, as shown in Figure 1.
To appreciate the likely evolution of AI-based automation, it’s important to evaluate the interaction of humans and machines across these five levels and to understand who analyzes the data, who decides based on the results of the analysis, and who acts based on the decision.
Source: IDC, 2018
In recent years, one of the shortcomings in the commercial sphere of AI has been the misrepresentation of the scope of possible automation. Too often, we hear claims of AI systems automating end-to-end processes and predictions of resulting massive labor losses. These proclamations or promises of the ability of AI to solve all societal ills, from diseases to crime and from hunger to war, do a disservice both to enterprises and individuals trying to plan for the appropriate level of investment in AI and to vendors developing marketing AI solutions.
For now, overwhelmingly, these claims are not substantiated as technical, organizational, financial, legal, and political barriers stand in the way of AI-based process automation. Therefore, the levels of AI-based automation must also be viewed in the context of the scope of automation. The scope of AI-based automation can be evaluated based on a hierarchy shown here.
Source: IDC, 2018
Why The AI Automation Framework Matters
This is a critical time in the evolution of AI development and adoption. The field, which has existed since the 1950s, is only now emerging as a viable commercial market. Many enterprises are placing bets on AI that will determine their future. Those that are sitting on the sidelines are risking being left behind. And yet many organizations lack the AI literacy needed to make critical investment decisions. The hype about AI is driving many decision makers that work with IDC to be either irrationally exuberant about AI or uncertain about the level of investments in data, algorithms, technology, and staff.
We believe our framework can be put to pragmatic use to identify industry and functional use cases where current AI can automate specific tasks, activities, or processes. It can also be used to better communicate to clients the value of AI capabilities through the lens of changing human-machine interactions and in the context of legal, ethical, and societal norms.
While business, IT, and analytics leaders need to recognize how AI is different from previous cycles of IT-based innovation, today’s leaders need to embrace AI and become involved in contributing to the discussion of AI ethics. Not only because a few can co-opt AI for nefarious purposes, but also because in the absence of human-driven ethical norms, commercial self-interest and technological evolution that incorporates emotional AI will likely lead to negative unintended consequences for your organization and society at large. With the broad participation of a diverse, global population in the conversation about the future of AI, we are more likely advanced through the levels of AI-based automation evolution while accumulating benefits for the largest possible population of humans.
For more information, see the complete Artificial Intelligence-based Automation Evolution Framework.