Artificial intelligence is spearheading the Fourth Industrial Revolution, the latest era of technological advancement. The growing appeal and utility of machine learning-based systems is undeniable. As Christina Pazzanese wrote in the Harvard Gazette in October 2020, “Worldwide business spending on AI is expected to hit $50 billion this year and $110 billion annually by 2024 … according to a forecast released in August by technology research firm IDC …. The company expects the media industry and federal and central governments will invest most heavily between 2018 and 2023 and predicts that AI will be ‘the disrupting influence changing entire industries over the next decade.'”
However, as AI becomes more prevalent, we need to consider its potential effects on society and organizations, as well as people. This article focusses on some of the challenges, including ethical, to consider when introducing ML- and AI-based systems inside your organization.
Machine Learning Challenges
As with any new technology, machine learning implementation brings challenges and risks that businesses need to face and mitigate before moving forward.
When you introduce a technology that provides judgment based purely on data, there is the risk of decisions being made which leave a large number of people behind. Automation can cause job losses, where the most likely candidates for automation are jobs that require repetitive manual tasks. The argument is often made that by using machine learning for repetitive tasks, it frees the workforce to focus on high order tasks. However, not everyone can be reassigned and reskilled for every job. As we move towards a future enabled by ML, we need to be aware of the challenges created as a result of increasing automation and provide for alternate job and revenue sources with reskilling.
Another challenge of an ML-based system is that it is heavily dependent on data rather than human insight. Such a system can easily lead to a “winner takes all market,” where large organizations with easy access to large quantities of data grow exponentially while the smaller players are left behind. A common characteristic with many of these technologies is the bigger they get, the better they get. Access to more data means better training for the systems, whereas a smaller player whose algorithms are just as good might not be able to match up. This could lead to a scenario of the rich getting richer and poor getting poorer in terms of data and growth models. We need to be thinking of pushing the edge of ML, with easy access to sanitized training sets and synthetic data for all.
Related Article: Ethical AI Is Our Responsibility
As organizations adopt ML-based systems, their ability to manage and adapt to this change is key to growth. As laid out in the steps for an ML implementation, the “change” in change management is crucial to ML success. Business processes and organizational structures often change as new technologies are adopted. It might even mean changes in the way we work, interact and govern our organizational structures. As ML-based systems become prevalent, we might even have our own paychecks decided by an algorithm that senses the market and compares that with our skill sets and sets our pay scale.
As Thomas Malone wrote in his book “The Future of Work,” new technologies will lead to more decentralized and matrixed organizations, spread across large geographical areas that use cloud-based video conferencing systems to come together. The crucial aspect of this technological change is to use the machine capabilities along with human capabilities to create an organization that can thrive and grow its people as well as culture.
Security is another issue related to machine learning.
Organizations and people are growing increasingly dependent on digital infrastructure for a range of processes, from payments to communication platforms. We now use ML-based systems to optimize business processes, improve productivity and enhance product quality. All of these systems are vulnerable to attack. So, as our dependency increases, so does our vulnerability where any vicious attack could cripple our ability to live and interact. As we have seen in the last few years, the ability of perpetrators to intrude and tamper with election systems, gas pipelines, health data and electrical grids could cripple not just an organization but entire countries and their economies. Integrating security and using ML-trained bots to actually predict, recognize and prevent attacks is crucial as ML-based systems become integral to our organizations.
Related Article: So, You Think Your AI Deployment Is Secure?
In an age where data is generated and tracked in so many different ways, privacy is becoming a huge challenge to address. We are already seeing companies track and use our shopping, viewing, traveling and eating patterns in the commercial market space. With the amount of personal data being continually created, ML algorithms that analyze the data can pose a significant challenge to privacy. For example, Amazon tracks and provides suggestions based on our activity. However, these same systems can also be used to track our personal health data, our relationships and financial information. Organizations need to find the balance between protecting privacy while not stifling the business growth and imperatives of ML-based systems.
Now Is the Time to Prepare for Machine Learning
Any technology disruption brings its own unique set of challenges, opportunities and risks. The ability of ML-based systems to overtake our own human insights leaves us even more vulnerable to vicious external attacks as well as an invasion to our privacy. The key is to recognize the socioeconomic, organizational as well as technical implications and brace ourselves for the upcoming changes. It may not be completely possible to make definitive predictions of how ML will evolve. In order to best prepare for the potential impact of machine learning on business and society, it is critical to first recognize the challenges and then manage the transformations through a dedicated change management system.
Author note: This is the third article in a series: In part one, we defined machine learning, delved further into the various types of machine learning models, and described their common applications. Part two was devoted to the VDCOR framework for implementing machine learning in your organization.
Geetika Tandon is a senior director at Booz Allen Hamilton, a management and technology consulting firm. She was born in Delhi, India, holds a Bachelors in architecture from Delhi University, a Masters in architecture from the University of Southern California and a Masters in computer science from the University of California Santa Barbara.
The views and opinions expressed in these articles are those of the author and do not necessarily reflect the official policy or position of her employer.