Artificial Intelligence and Leadership Development

Working in the fields of both leadership development and technology, I track where the two intersect, and where new tools and technologies are likely to have a significant impact on leadership and coaching. Here are some of my observations and thoughts on how artificial intelligence-enabled technologies have already started impacting our work in leadership development, and where it’s headed.

Sentiment and Thematic Analysis

Early commercial uses of artificial intelligence (AI) have been in the area of natural language processing (NLP) which has been used for sentiment analysis — the ability to determine the emotional meaning of unstructured text. As machine learning has become more sophisticated, tools have now been developed that can perform thematic analysis as well as sentiment analysis. 

This has rapidly become an essential tool in analyzing large amounts of survey responses and automating the task of reading and coding text, which previously wasn’t cost-effective at scale. The demand for efficiency in market research has driven the commercial development of survey platforms, which have also served organizations working in employee engagement and organizational culture. This is now a mature technology ubiquitous in survey research and is the same technology used in chatbots — the subject of continuous research and development.

When we look at developing leaders specifically, rather than understanding the organization more generally, we move to individual and team assessments as data collection tools. Here, the data set is much smaller, so the benefits of automation are much less. For example, a human can still read comments from a 360 assessment and understand the themes in a more nuanced way than thematic analysis software (at least right now).

Benefits for leadership development are likely to come from either:

1. Examining much larger dataset for an individual, e.g., analyzing communications across email, video and social networks — data gathered as a part of “getting work done” rather than from assessments, or 
2. From AI reaching the point where it’s providing insights from a smaller dataset that a human wouldn’t necessarily reach by analyzing the same data. 

This level of insight from machine learning algorithms is going to require large sets of executive-level data and an evaluation of exactly what effective performance looks like. This is very hard to do holistically — to ask whether someone is an effective leader — particularly as there is no single objective measure of leadership effectiveness. It is dependent on many factors, whose weighting is subject to personal opinion.

 But there are certainly areas where big data analysis and AI tools will be able to help. There is an entire field of research on social influence, and tools exist that can analyze corporate communications to identify employee influence across the organization, which is certainly a factor in effective leadership. Another area is quality of communication — but more on that later. A future where we can bring the data together from these different types of analyses and then use AI to mine for insights is ahead of us, but it’s still early days in this arena.  

Intelligent Matching Algorithms

The next major use of AI more generally in talent management has been in skills search and matching. Recruiting platforms (of which there are now hundreds) have driven this through resume-to-job matching. Recruiting is where the bulk of HR-related innovation investment goes due to the huge potential to reduce recruiting costs. These innovations have then been leveraged for internal organization use for project staffing, internal job search, and for mentoring and skills coaching, with companies such as eightfold.ai built from the ground up as talent AI platforms.  

Use of artificial intelligence in this context though is in the “red flag” usage area of AI. As soon as we provide advancement opportunities and base them on computer algorithms, we bring in the potential of unintended bias, diversity issues and employment law violations. How we identify, manage and regulate these issues is going to be a major societal issue in the coming years, and how that will put the brakes on innovation and adoption is yet to be seen.

As more organizations start to use matching algorithms in coaching and mentoring software, we also run into a fundamental issue around the purpose of this type of development and what it means to match individuals who can contribute to someone’s development.

If the purpose is developing a specific skill or changing a specific behavior, then connecting to someone who is an expert in that area, or expert at coaching to that need, makes sense. Similarly, with mentoring, we are often trying to match an employee with someone who has already “trodden the path” that they wish to follow, and you can look for someone who has higher order capabilities in the same field.

However, once we move into leadership development, and particularly executive development, more frequently we are aiming to shift the person’s mindset — their ways of thinking about and responding to situations. This requires exposure to different perspectives and approaches, so there’s often a balance between having close enough experiences to understand the leader’s environment, but far enough away to experience other industries, functions, cultures, and challenges.  Technology tends to reinforce the idea of finding similarities, and surfacing “people like you,” which is the opposite of what we are trying to do. This is potentially a solvable technology problem, but a challenging one that I haven’t seen addressed in this context to date.

Generative AI

2023 was the year of ChatGPT when generalized natural language AI became a tool for everyone. A lot of the data for individual leadership development comes from interviews and assessments, and I see ChatGPT already being used to summarize notes and analyze themes in survey responses and interview transcripts. The specific case I mentioned earlier of whether a computer can analyze themes in a feedback report better than a human is reaching a tipping point. A year ago the answer was no, but by next year it may be yes. A better answer may be that a human using AI as a tool to support them will provide a better-quality analysis than either one on their own.

If you ask ChatGPT how it benefits leadership development, it will tell you it can provide role-playing scenarios and curate leadership development content. Currently, ChatGPT, along with creative generative AI tools such as Runway for image generation are great tools to augment human design capabilities in these areas. These tools haven’t been available long enough to see more dynamic real-time use of the technology for nuanced leadership-level training, but it will be interesting to see how the training industry adopts AI technologies in the next few years.

For generative AI, I think we’re just at the beginning. Just as when we first realized computers could be used for more than just doing calculations, we have a whole world of uses for generative AI that are out there waiting to be discovered. We may still be a long way from ChatGPT’s own claim when asked that it can “provide leadership coaching anytime, anywhere due to its availability round the clock” — but, if we look back at our experience with innovative new technologies, the best future uses will likely be something no one has even thought of yet. 

Communications Analysis

While it may still take a human to conduct a holistic analysis of leadership capabilities, there are certainly specific skills where AI can provide development support. One of those is communication skills. The ability to communicate effectively with the organization and your team, and improve the way you give and receive feedback — particularly having development conversations with others — is a frequently identified growth area for leaders.

While the public focus recently has been on ChatGPT, AI product companies have taken basic sentiment analysis to a much more sophisticated level. Over the last few years, commercial tools such as Thematic and Dovetail have emerged that can take chat, conversations and survey data and transcribe and conduct thematic analyses on them for marketing, product and support insights. Likewise, companies such as SproutSocial are using AI for monitoring and managing public relations and other communications.

Products such as Poised and Yoodli not only analyze what you say, but how you say it. They analyze volume, cadence and pauses to evaluate confidence, energy and empathy, and they can identify how many questions you asked and were asked. This allows you to get feedback on any online conversation you have, and potentially help you improve your communications. This analysis is moving to real time, so you can receive feedback and adjust mid-conversation.

If you then layer the ability to interpret expressions and reactions via video analysis, you are really starting to analyze communication, which involves listening, questioning and emotional connection, and not just one-way speech. We’re starting to see tools for doing that, such as Ovida, which is marketed to coaching and leadership schools to help train coaching skills. They provide a videoconferencing platform that analyzes your conversations, including the back and forth between individuals, identifying how much you speak vs listen, and how much you pause and ask clarifying questions.

We know that receiving regular reminders and constructive feedback are critical to adopting new behaviors. These new tools have the ability to provide continuous feedback in every video interaction. 

It’s easy to imagine a not-too-distant future where, once proven, these technologies become ubiquitous. You could envision Microsoft acquiring one of these innovators, scaling it out on their global cloud platform and embedding it in Microsoft Teams so that millions of business communications are being analyzed in real time and individuals receive feedback on every online interaction. Microsoft is already rolling out Copilot, their AI assistant into Teams and Office, so this would just be a logical upgrade to those capabilities.

Like most AI applications, there are ethical, legal and security considerations — but if individuals’ own communications are being analyzed and fed back to only them, this is a reasonably low-risk application of AI that could spread quickly.

The more dystopian view would be of organizations continuously monitoring the communications of employees for changes in patterns of communications in response to external events or internal communications, allowing them to carefully craft messaging to influence and control their employees.

However, I’m optimistic we can manage the use of AI and get the balance right, and if AI can help us all communicate better and have more open and productive conversations with one another, then maybe technology can help the world become a better place to live and work.