How Leaders Are Improving Their Organizations' AI Adoption and Performance

AI is already changing how organizations and the people in them work. But rather than the feared wholesale replacement of entire categories of employees, many leaders are finding that AI is causing the reinvention of how people work. 

Successful leaders are finding that learning, inclusion, breaking down of silos (especially between IT and other business functions) and recognition are the keys to optimizing AI adoption and performance. JL 

David De Cremer reports in Harvard Business Review:

Business leaders don’t have to be AI experts. They need to be AI-savvy enough to recognize (its) benefits for the organization and its stakeholders. Leaders must learn to empower human-AI collaborations to identify opportunities for AI integration in everyday workflows. AI cannot entirely replace humans and cannot think for us. Savvy leaders prioritize participation. Successful human-AI collaborations cross disciplines (between tech and business functions). Different teams feel more or less threat (or benefit) from AI, (so) may turn to siloed behavior, become averse to working with AI, never develop trust in its capabilities, and resist even positive changes that come from using it. Employees are crucial to the performance of AI and deserve acknowledgment the value AI creates

AI is intimidating your employees. As machines increasingly perform intellectually demanding tasks that were previously reserved for humans, your people feel more excluded and less necessary than ever. And the problem is getting worse. According to the market research company Vanson Bourne, 80% of organizations say that their main technological goal is hyperautomation—the end-to-end automation of as many business processes as possible. Executives have a tendency to pursue that goal without any feedback from their employees—the people whose jobs, and lives, will be most affected by achieving it. But my decades of research into the enterprise adoption of emerging technologies has proved one thing time and again: The savviest leaders prioritize participation by the rank and file throughout the adoption process.

When employees are excluded from that process, they become averse to working with AI, never develop trust in its capabilities, and resist even the positive changes that come from using it. Nonetheless, done correctly, human-AI collaborations represent the most promising way of working. They may not always be the fastest, cheapest, or easiest way to introduce and use artificial intelligence, but the alternative, which excludes workers, is no alternative at all. Consider one example, from researchers at New York University’s Center for Cybersecurity. The research team used Copilot, a tool developed by GitHub to generate code automatically, to produce 1,692 software programs with no input from human coders. Forty percent of those programs had critical security flaws.

In this article I examine what keeps leaders from including rank-and-file employees in AI projects, how they should model inclusive behavior, and what your organization must do to develop employee-inclusive AI practices. Those practices can make your long-term performance more likely to improve and your employees more likely to be happy, productive, and engaged.

Becoming Comfortable with AI

You can’t bring everyone into the AI adoption process if you’re not heavily involved yourself. But business leaders often ask me how they can guide an AI-based transformation when they have no personal expertise with the technology.

Business leaders don’t have to be AI experts. They only need to be AI-savvy enough to recognize the technology’s benefits for the organization and its stakeholders. Once AI has been deployed, leaders must learn to empower and drive human-AI collaborations. For example, they should be able to identify opportunities for AI integration in everyday workflows and to anticipate its potential advantages for teams and projects associated with the technology. In short, learning must be part of their ongoing AI leadership.

Some executives in my advanced leadership classes have wondered aloud whether they need to become professional coders to be effective leaders. What they need is not coding expertise but a foundational understanding of the technology.

The Basics of AI

Most managers know that AI tools are computational systems that have autonomous learning ability. They understand that AI can learn from large datasets and engage in pattern recognition and problem-solving. They’ve probably already seen it used in a variety of organizational applications: scanning the résumés of job applicants, evaluating employee performance, optimizing task scheduling, managing inventory, and automating repetitive tasks so that employees can explore new ideas and promote innovation rather than count widgets. It’s AI’s ability to learn—using algorithms to process new data and change its computation of information based on that data—that results in comparisons to human intelligence. But too many business leaders implicitly assume that AI can take over almost any position from humans.

The reality is that AI cannot think like a human, and it isn’t all that creative. First, it generates no novel ideas; its ideas exist in the datasets that are fed into it. Not even the most sophisticated AI systems can infer meaning from learning, as humans do. They cannot draw analogies, and they cannot appreciate cultural and contextual nuances. Whereas humans can extract the deeper meanings and intricate nuances of business conversations, AI cannot tell when what is said is contradictory to what is meant. For example, it will interpret “You’re serious about this offer?” as a simple request to confirm what is being offered. Most humans will understand that the other party is unhappy with what is being offered.

Business leaders who are just AI-savvy enough recognize that the technology can do much to improve work efficiency and the overall functioning of an organization. They must also recognize that it cannot entirely replace humans and, most important, it cannot do our thinking for us.

Three Ways AI Can Alienate Employees

Once you’ve become comfortable with your ability to discuss and champion AI adoption, you’ll need to generate enthusiasm throughout the rank and file—not an easy process. To be an effective leader, you must understand why AI causes a rift between your workers and management and find ways to bridge the gap between what they’re feeling about it and what you’d like them to feel. And you’ll need to prevent the territorialism and tribalism that can occur when one group controls AI and another doesn’t even understand it.

Here are three common reasons for workers’ alienation.

Employees lose autonomy and become cynical.

 

Not long ago a colleague of mine applied for a credit card at her bank. The employee helping her entered all her information into a computer program, which ran an algorithm to determine whether she qualified. My colleague, who earns a good living and has good credit, was surprised when the employee informed her that the algorithm had decided she did not qualify for the card. When she asked for an explanation, he replied that the decision was fact-based and automated, so he could not add much to it. Eventually he mumbled that he was not a machine, so why should she expect him to understand the algorithm’s decision? That comment revealed that the employee did not feel in control of his job, was clearly demotivated, and had no intention of trying to make the algorithm’s decision comprehensible to my colleague. The result was poor customer service and a missed business opportunity.

When you automate easy tasks but leave difficult and emotionally demanding ones to humans, you negatively affect the well-being of your workers. A 2021 study from Georgia State University revealed that the more automation is introduced in the workplace, the worse employee health and job satisfaction become.

Employees don’t understand AI and resist it.

 

People generally prefer to work with and receive advice from humans rather than AI. You should be aware of this bias and recognize that employees will respond emotionally rather than rationally to the technology—even when it has proved to be superior to humans.

If you want to make AI adoption inclusive, you must position yourself as both a mediator and a facilitator in human-AI interactions. You need to ensure that your employees receive adequate support and training to interact effectively with AI systems and to create opportunities for them to turn to a human if those interactions go wrong. If they feel truly included in how you plan to work with AI, they will be less averse to it.

Chad Hagen

A failure to be inclusive may even lead to active resistance. For example, when workers at Amazon’s packing facilities were “supervised” by AI algorithms, they became more injury-prone. They were forced to meet high productivity targets, with few if any opportunities to take a break, and could be indiscriminately fired for not hitting their targets. Frustrated, they signed petitions and gathered outside their warehouses, united by the rallying cry “We are not robots!” Indeed, as one employee succinctly put it, “[Productivity is] all they care about. They don’t care about their employees. They care more about the robots than they care about the employees.”

If you want to avoid resistance from your employees when introducing AI, you must push them out of their comfort zone while ensuring that they understand why you’re doing so. They should know how you plan to take care of them during this transition. You’ll need to exercise patience, because it will take time and effort for workers to become familiar with AI and see how it can help them in their jobs.

AI creates business silos.

 

In addition to eliciting resistance, AI adoption can undermine inclusiveness by entrenching silos in your organization in three ways. First, because the deep expertise required to understand and operate AI systems is often found only in tech teams, employees in other departments (such as HR, operations, and marketing) may have difficulty interacting with AI. But they need the know-how to make use of it in ways that are meaningful to their own business goals. Second, data ownership and access can be a contentious issue between departments. AI systems rely heavily on data for training and decision-making, but individual teams may have their own data repositories and be unwilling or unable to share data with others. Third, the impact AI has will vary across teams: Some may find it more useful than others do, and some may see it being used to automate their tasks more than the tasks in other departments. When different teams feel more or less threat (or benefit) from the adoption of AI, they may turn to siloed behavior, avoiding collaboration and information sharing to protect their own interests.

Employee resistance often creates an organization in which experts in AI and those in business work separately. People mentally shut down and live within the realm of their own expertise. And when AI is adopted differently across silos, resources may be duplicated or underutilized, limiting leaders’ ability to scale up the technology across the organization. Teams may collect, store, and manage data independently, resulting in inconsistencies, redundancy, or incomplete datasets. That can hinder your ability to leverage the full potential of your data. When departments operate in isolation, cross-functional collaboration and interdisciplinary problem-solving become impossible. It will be your job as an inclusive leader to stress the importance of collaboration and push for the implementation of technological and organizational solutions, such as centralizing data for analysis in cloud-based tools.

To address all these challenges, you need to adjust your organization’s culture.

A More Effective and Inclusive Model for AI

As a business leader, you have to make people feel like full-fledged members of your organization—empowered to work like human beings while collaborating with AI in every automated process. AI can quickly produce code for new programs, for example, but human employees are needed to fix any security flaws and other glitches.

An inclusive approach will make employees feel in control of the adoption process, reduce aversion to the technology, and increase trust in it. Those outcomes will help integrate it more efficiently in your employees’ workflow and will enhance the likelihood of creating value across the organization (rather than establishing only siloed, and thus minor, effects). To achieve them, the organization must consistently follow four practices.

Create space and time for social connection.

 

When working with AI, people have to spend a lot of time in front of computer screens communicating with machines. That limits their interaction with other humans. A 2022 poll by the Pew Research Center revealed that a major concern people have about the presence of artificial intelligence in their lives is that it isolates them from other humans. As a leader, you have an important responsibility to foster the social connections of your employees, which you can do through events and online communities within and outside the organization. Digital underwriters, for example, often issue insurance policies without even meeting applicants. They could be asked to have weekly meetings with other underwriters and with the people who built the AI system they use to discuss possible improvements. Uber now allows its drivers, who are under constant algorithmic supervision and feel dehumanized as a result, to telephone other people in the organization when they need help or have a question.

The Fortune 500 dairy company Land O’Lakes provides an excellent example of how to free employees from the solitude of working with AI. It began its AI transformation in 2017, when it sought to partially automate commodity forecasting and propensity modeling. Company leaders prioritized speaking with the rank and file about the expected challenges, helping establish a common understanding of the project’s possibilities and limits and assuring people that the company wasn’t pursuing tech for the sake of tech. Teams coordinated across departments, but company leaders also conducted weekly people-to-people check-ins with every business unit to address any challenges, emotional or procedural, that may have arisen. That approach was crucial to the success of Land O’Lakes’ AI transformation. Employees were given opportunities to voice concern, to question tactics, and to raise anything else that might be on their minds.

Make tech and nontech teams collaborate.

 

As an AI-savvy leader, you know that successful human-AI collaborations cross disciplines. Your tech and business experts should not retreat to their separate corners, literal and virtual. So build diverse teams that work together to adopt AI. For example, business experts can explain to tech experts what goals must be achieved, and tech experts can make suggestions regarding which AI systems will be needed. Meanwhile, HR can familiarize employees with the AI system they’ll be using and the skills they’ll require, and operational staffers can focus on integrating the entire human-AI workflow into the organizational setup.

To lead such diverse teams and bring them together, you must communicate in ways that unite rather than divide people, allowing for and integrating multiple perspectives and identifying roadblocks that may complicate or prevent collaboration. As a business leader, you can start by explaining the organization’s needs to your tech and business teams and then outline how the tech experts will become part of the business process to achieve the desired results. Try to establish a common language and understanding for both groups regarding how to approach challenges, recognize patterns, break big problems down into smaller ones, and find a shared work method. Without that common language, your teams may fail to cohere, and the inclusive culture you’ve tried to develop may dissipate.

In one of my consulting projects I watched the chief technology officer of a global financial institution present the company’s new tech strategy. Just a few minutes in, the CEO interrupted. He said he didn’t understand anything the CTO was saying and pressured him to present his message in three simple bullet points. It was embarrassing for the CTO. The tech team retrenched. IT departments stopped trying to talk to top executives. The CEO lost credibility with senior executives, who realized he wouldn’t be capable of guiding the bank through its AI-adoption project. He hadn’t become AI-savvy, didn’t connect AI to the purpose of the company, and, worst of all, had not developed the inclusive mindset needed to translate from the CTO to the business and back. Needless to say, the project failed. The CEO left the company the following year.

Successful human-AI collaborations cross disciplines. Your tech and business experts should not retreat to their separate corners, literal and virtual.

When done properly, mixing teams can fundamentally improve not only a company’s technology but also its overall culture. In 2017 the agricultural equipment maker CNH Industrial’s leadership team decided it wanted to create a host of AI-powered automation capabilities. It also wanted to connect customers with internal and external partners and promote CNH as a service-oriented business.

The executives began the transformation process by speaking with employees from its commercial vehicle unit, industry-specific vehicle units, IT, and operations. Digital advisers and a new digital team were created within CNH’s existing IT organization to support ongoing strategy, implementation, and execution. By establishing cross-disciplinary teams and keeping them involved throughout the process, CNH was able to quickly adopt (or retire) experimental approaches. It lowered the barriers between developers and business owners, and it allowed for real-time feedback on scheduled work.

Constantly develop your own leadership skills.

 

Making your employees feel included in your AI adoption project requires that you account for their uncertainty and discomfort when dealing with AI. As an AI-savvy leader, you should be seen as open to listening to their concerns. My research indicates that employees are indeed more willing to trust and engage with AI if their leaders are humble and demonstrate that openness.

Consider Satya Nadella, the CEO of Microsoft, who is a master at using empathy to foster inclusion. One of the first things he did when he was appointed CEO, in 2014, was to persuade his employees that no matter how successful Microsoft had been in the past, they should stay open to new ideas and other ways of working. Asking them to think differently required courage, but it also showed the importance of being humble—unafraid of receiving feedback from others. A humble attitude in a leader encourages employees to interact regularly with experts in different departments to understand and relate to the diverse perspectives at work in the organization.

 

You must also guide employees in their understanding of AI. For human-AI interactions to be truly collaborative, employees need strong frameworks for thinking about how to work with smart machines. In airline safety, for example, pilots need more training to fly planes with collaborative autopilot systems. That’s because, as Captain Shem Malmquist, a veteran safety and aviation accident investigator, told Wired in 2022, they “must have a mental model of both the aircraft and its primary systems, as well as how the flight automation works” to manage issues that could turn into catastrophic crashes. Only when employees have a clear model of their own strengths and weaknesses, and those of their AI tools, will they understand how AI can augment their work.

Reward workers for being human.

 

Employees want you to tell them how you see their role in the human-AI collaborative process. They also want to know how they will be rewarded for the value that collaboration creates. For humans and AI to work together successfully, you need to establish clear guidelines for who is credited with what. Otherwise your employees may feel that you’ve downplayed their contribution and attributed the project’s success largely to the AI.

To ensure that employees feel included, let them share in the rewards that come with the value that AI creates. Emphasize that in your view, humans are crucial to the performance of AI and therefore deserve appropriate acknowledgment. Even just a companywide email recognizing and celebrating someone’s accomplishments can go a long way toward boosting morale.

. . .

AI adoption is a complex process that requires everyone involved to learn, question, and collaborate. How your company approaches it will depend on the level of your employees’ technological acumen, your budget, and many other critical factors. But the approach I recommend is one that any company can take to optimize the process.

It should begin with managers’ learning just enough about AI to feel confident communicating its importance to their teams. Then you need constant human-to-human connection among cross-disciplinary business units as well as meetings at which everyone feels free to speak openly. Such gatherings provide excellent opportunities for managers to show vulnerability, communicate their own questions, or even just listen to venting among colleagues. When your transformation is underway, and your business is focused on optimizing AI rather than simply implementing it, you should reward your employees for their uniquely human contributions. If they don’t feel valued and respected, your transformation attempt will certainly fail.