AI Works. Now We Need to Make It Work for Us.
In an article recently published in Fast Company Magazine, author John Paylus describes the amazingly rapid pace at which artificial intelligence (AI) has made it from the laboratory into the Western industrial workplace. Compared to the Internet — which took about 20 years to become widely used — and smartphones — which took about 10 years to become ubiquitous — AI took less than a decade from the development of its modern form in 2012 to contributing $2 trillion to the worldwide economy by 2020, as estimated by PriceWaterhouseCooper. (By the way, that’s trillion, with a “T.”)
For Paylus, this is good news… and bad news. The bad news is that, like all new technologies, AI is bringing with it a raft of problems. For example, in-real world applications, AI-based face recognition software has shown itself to be rather, well, racist and sexist, being able to easily distinguish features of individual white males but having clear difficulty with non-whites and women. (Perhaps reflecting the bias of its designers.)
Meanwhile, so-called “deepfakes” — hyper-realistic computer-generated videos showing people doing and saying things they never did or say —threaten to disrupt political discourse, especially on social media. The good news? AI adaptation is happening so quickly its faults are being recognized in real time. Unlike the pollution and the climate-threatening impact of automobiles, which took decades to recognize, or the alienating impact of cell phones, which took years to be identified and understood, the rapid adoption of AI is making its negative impacts stand out in stark relief. And being able to more easily recognize the problems means we’re also able to more quickly mitigate their effects.
So, how can today’s CEOs who are integrating AI into their workplaces help ensure the technology performs as advertised? And how can they enjoy the benefits of artificial intelligence while mitigating its negative effects?
In Reinventing Jobs: A 4-Step Approach to Applying Automation to Work (Harvard Business Press Review, 2018) authors Ravin Jesuthasan and John W. Boudreau, both consultants with Willis Towers Watson, provide an excellent framework for best integrating AI and other new technologies into the workplace. Their primary message: This isn’t a binary choice. It’s not Yes or No. All or Nothing. Instead, the goal is to create ideal human-machine combinations.
Here are the four distinct steps Jesuthasan and Boudreau suggest:
1. Deconstruct. The first step involves deconstructing jobs or workflows into component tasks and categorizing them along such continuums as Repetitive vs. Variable, Independent vs. Interactive, and Mental vs. Physical. Repetitive jobs are more easily automated than variable ones. Tasks that can be performed independently are more easily automated than those that require many parties to work together. And physical tasks, like lifting, moving, or assembling, are easier to automate than those requiring thought, analysis, judgment, and creativity.
2. Optimize. Once a company has identified the tasks most suited to automation, its goal should be to identify how matching humans with machines/AI can improve overall work outcomes. For example, a company could use RPA (Robotic Process Automation) for repetitive mental tasks to help reduce mistakes. Or employ Social Robotics to augment repetitive physical tasks to ensure consistency.
3. Automate. Jesuthansen and Boudreau identify three ways one can automate a task or process:
Robotic Process Automation (RPA): This is best for high-volume, low-complexity, and routine tasks, such as assembly line work or moving objects from one point to another. It can also be used for organizing data, such as transferring info between software systems or using simple rules to find content in emails or spreadsheets and entering them into business systems, like enterprise resource planning (ERP) or customer relationship management (CRM).
Cognitive Automation: This employs tools, like pattern recognition and language understanding. Combined with machine learning, this innovation has been used to draw insights from Big Data and perform diagnostic tasks, such as identifying cancers in X-rays.
Social Robotics: Here, the word “social” refers to robots that move around and interact with people, using sensors, AI, and mechanical machinery. A subset of social robotics is “collaborative” robotics (cobots). Cobots are machines that actually sense the human worker and actively adjust to physically work with the human.
4. Reconfigure. Once a job has been automated, it can change the job itself in one of three ways. Either the robot/AI can substitute for the human working, rendering that individual obsolete; it can augment the human worker, allowing him/her to do more, do it better, and/or do it in less time than before; or it can transform the task entirely, making it fill a completely different role than prior to pre-automation.
In his 2019 best-seller Talking to Strangers, Canadian journalist Malcolm Gladwell explores how human beings are biologically, psychologically, and emotionally ill-equipped to properly judge the characters, motives, and behavior of other human beings, especially those with whom they are not already well-acquainted. People are intrinsically handicapped by their own prejudices, biases, and personal experiences to render objective judgement on people of other cultures, ethnicities, races, social classes, and other demographic distinctions.
Currently, AI has been shown to share similar failings, as it is the product of programmers who necessarily share these same flaws. But, as John Paylus suggests, now that we recognize this issue, we can move to correct it, allowing AI to deliver the benefits to transform our workplace for the better as Jesuthansen and Boudreau describe.
For more information on how I can prepare your business to get the most from the AI revolution — while avoiding pitfalls in the ongoing talent war, please read an article profiling me in Forbes. And if you want to discuss how you can get your organization up to speed in the modern era, email me @ laura@conoverconsulting.com. You don’t need AI to see the time to act is now!