We Replaced Leaders with Artificial Intelligence. Here’s What We Learned:
by Mengqiao Liu, Ph.D.
Five million jobs are projected to be replaced by technology by 2020, and, according to the 2016 World Economic Forum, 50 percent of today’s workforce activities will be automated by 2055. Artificial Intelligence (AI), the study of machines that can imitate human intelligence, is undeniably revolutionizing our world, transforming the way we learn, shop, entertain, and more.
Immersed in this AI era, you’ve probably wondered if you should start automating your organization, as well. And, if you do, you may be wondering if there will still be a place for leaders in the workplace in 20 years.
We wondered the same. To find evidence-based answers, we started experimenting with AI and applying it to leadership selection and development. Our conclusion? Leaders are not only here to stay but are becoming more important in the era of AI.
Here’s how we arrived at this conclusion. We started our AI venture in partnership with MedRespond, a company that combines AI, search, and media to provide training solutions. Leveraging natural language processing and machine learning, we developed an interviewing simulation where leaders can practice interviewing skills. This powerful tool can train leaders to ask behavioral questions to elicit job-relevant information, as well as deter them from asking biased or illegal questions (e.g., “How old are you?”). Given that interviewing requires specific, focused, procedural skills that can be captured using data, AI was quite effective in this case, achieving over 90 percent accuracy.
Where AI falls short
In another AI experiment, we explored whether AI can be used to interpret and evaluate leadership behaviors. But after analyzing data on thousands of leaders’ interactions with employees and coworkers, we found that AI falls short in understanding important leadership behaviors, such as coaching and delegation, as it fails to differentiate effective leadership behaviors from ineffective ones.
Let’s take coaching as an example. Our machine learning algorithms were given the task to dissect all the key behavioral actions that take place in a coaching conversation (engages the employee, defines challenges, offers support, gains commitment, etc.) and evaluate whether each action is effective—which, by the way, even a human leadership specialist may find challenging to do from time to time. It’s expected that computers would struggle at these tasks where complex and multidimensional knowledge, skills, and interpersonal understanding are required. So, score one for humans. After all, if AI can’t tell the good from the bad, how can we expect it to successfully lead people?
Taken together, our findings reflect the strengths as well as the limitations of today’s AI, which focuses on solving for narrow tasks that can be done with less than a second of thought by a human. While AI can take on specific, narrow-focused tasks that can make a leader’s job more efficient (e.g., using data trend visualization to aid decision making), leadership requires more than AI can offer.
Leadership is complex and multidimensional. It has to do with communication, empathy, and trust, all of which, in addition to complex knowledge and skills, cannot be mastered by AI now or in the foreseeable future. Therefore, AI should be seen and leveraged as a supplement to a leader’s job and not a substitute for it.
How to best use AI
Here are our recommendations for leaders deploying AI within their organizations:
Interested in learning more? Read about how DDI can help you develop and transform leaders at all levels.
EN KISA SÜREDE
SİZİNLE İLETİŞİME GEÇECEĞİZ