Artificial intelligence is changing the way organizations serve customers, support employees, and make decisions. Nearly every week brings another announcement about faster models, greater automation, or new applications that promise to improve productivity and personalize experiences. It is easy to become captivated by questions about what AI can do. 

​

Yet I find myself asking a different question. 

​

What happens when leaders don't see the consequences of what AI is doing? 

 

That question is not really about technology. It is about leadership. 

 

Every leadership decision carries two responsibilities. The first is making the decision itself. The second is creating a way to learn whether that decision produced the intended outcome. Organizations have always depended on feedback to improve products, strengthen relationships, and adapt to changing environments. Peter Senge described this as one of the defining characteristics of a learning organization: the ability to recognize patterns, understand feedback loops, and adapt before problems become crises (Senge, 2006). 

 

One of leadership's greatest challenges, however, is that consequences rarely arrive on the same day as the decision. This is a characteristic of complex social systems found in many organizations. Donella Meadows describes these as delayed feedback loops where the effects of an action are not observed until some time after the action occurs.  

• We launch a new policy in January and see employee turnover in June. 

• We reduce staffing in March and hear customer frustration in August. 

• We introduce new performance incentives and discover a year later that we unintentionally changed the culture. 

​

A classic customer experience example is leadership reduces onboarding, coaching, and ongoing training to lower costs to meet budget targets.  

​

The early indicators look encouraging. Training costs decline. New employees reach the floor more quickly. Productivity appears stable because experienced employees compensate for knowledge gaps. 

​

Months and even years later, however, a different story begins to emerge. Errors increase. First-call resolution declines. Repeat contacts rise. Supervisors spend more time correcting mistakes, and experienced employees leave, taking organizational knowledge with them. 

​

The original decision wasn't necessarily wrong. The organization simply mistook early efficiency for long-term effectiveness. 

 

Leadership has always operated within delayed feedback loops. Artificial intelligence has not changed that reality. It has simply accelerated the speed at which decisions are made while increasing the distance between those decisions and their consequences. 

 

For years I have been an advocate for artificial intelligence because I believe it has the potential to make us more human, not less. I believe this because AI excels at work humans were never meant to spend their lives doing. It recognizes patterns across enormous datasets, automates repetitive tasks, detects anomalies, predicts future behavior, and reduces friction in everyday experiences. Customers have quietly embraced these capabilities because they make life easier. 

Every day we trust AI to recommend products on Amazon, reroute us around traffic through Waze, detect fraudulent credit card activity, remind us to refill prescriptions, schedule appointments, and personalize streaming services. Most people rarely think about the technology because the experience feels effortless. 

 

Much of today's questions about AI focuses on capability. 

​

• Can it automate? 

• Can it summarize? 

• Can it predict? 

• Can it replace manual work? 

​

These are worthwhile questions. The opportunity for organizations to become more human is through answering a different set of questions.  

​

• What relationship are we trying to build? 

• What promise are we making to customers? 

• How will people experience the decisions our systems make? 

• What happens when no one notices something is going wrong? 

​

Those questions move the conversation from technology to leadership. 

 

Technology recognizes patterns. 

Customer understanding gives those patterns meaning. 

Leadership determines whether those patterns should influence decisions. 

 

Without all three working together, AI becomes an extraordinarily efficient pattern-recognition engine making decisions disconnected from the human outcomes organizations hope to achieve. 

 

This distinction matters because AI does not actually understand customers. It recognizes statistical relationships, predicts probabilities from historical data, and identifies similarities across millions of observations. Those capabilities are remarkable, but they are fundamentally different from understanding what matters to people. AI can identify patterns in behavior. It cannot determine purpose, values, or the meaning customers assign to an experience. That remains the work of leadership (Russell & Norvig, 2021). 

 

Essentially, customers continue to seek something AI cannot fully provide: explanation, judgment, empathy, and accountability when a decision feels unfair or difficult to understand and responding to that need requires context. It requires knowing what matters to people, what they are trying to accomplish, what obstacles they face, and how their experiences shape trust over time. 

 

Organizations often describe this through customer relationship management, but CRM is much more than software. At its best, it represents an organization's collective understanding of the relationships it hopes to build. Recent research demonstrates that AI-enabled CRM systems can improve customer retention, personalize experiences, predict customer churn, and enhance operational performance. Yet those same studies conclude that these outcomes depend on ethical design, transparency, and a deep understanding of customer relationships—not simply better algorithms (Rainy & Goswami, 2025). 

 

This is why recent litigation involving AI-assisted hiring platforms such as Workday deserves our attention—not because it is ultimately about legal liability, but because it raises a broader leadership question. 

​

If no human reviewed the decision, who learned from it? 

I am less interested in how courts answer that question than I am in how organizations answer it. Delegating portions of decision-making to AI does not delegate accountability. 

 

Let’s be clear, the risk is not that AI makes decisions. The risk is that, with AI, organizations become exceptionally good at making decisions before they become equally capable of learning from them. As technology becomes more sophisticated, leadership must become equally sophisticated in learning. 

 

Peter Senge reminded us decades ago that learning organizations distinguish themselves by their ability to recognize patterns, surface assumptions, and adapt before problems become crises. Artificial intelligence dramatically expands our ability to recognize patterns, but it does not automatically make organizations better learners. That responsibility still belongs to leadership. 

 

Leadership's challenge, therefore, is no longer simply selecting AI. It is designing feedback systems capable of learning at the same speed AI makes decisions. That means continually asking difficult questions. 

• What is our AI actually producing? 

• Who benefits? 

• Who may be unintentionally excluded? 

• What assumptions are embedded within our data? 

• What populations were underrepresented during design? 

• Which customer experiences are improving? 

• Which ones are quietly deteriorating? 

 

Organizations must also recognize that context matters. Sector, geography, demographics, regulation, and culture all influence how technology is experienced. A model that performs exceptionally well in one environment may produce unintended consequences in another. One algorithm cannot be expected to produce equitable outcomes for every population under every circumstance (Rainy & Goswami, 2025). 

The encouraging news is that organizations, like USAA, Costco, and Cambia, that genuinely understand their customers already possess a tremendous advantage. 

 

Artificial intelligence, when thoughtfully governed by leadership, can personalize experiences, detect customer churn earlier, identify emerging needs, reduce customer effort, support employees with contextual insights, and improve service consistency. Organizations can utilize AI to strengthen trust and organizational performance when combined with thoughtful governance, transparency, and human oversight. However, technology alone does not create trust. 

 

Trust emerges when customers believe organizations understand them, explain decisions transparently, learn from mistakes, and continually improve the experiences they create through automation and human connection. Elevating customer experience has never been about economizing or eliminating human interaction. It has always been about eliminating unnecessary human effort while preserving meaningful human connection. 

 

Perhaps this is the greatest opportunity AI offers leaders. Not simply to automate work but to become better learners.  

• To hear weak signals before they become crises. 

• To identify unintended consequences before they become lawsuits. 

• To understand customers before they become dissatisfied. 

• To improve systems before someone else points out what should have been obvious. 

 

Artificial intelligence will undoubtedly become more capable, and that is a good thing. The question confronting today's leaders is not whether AI will continue to evolve. It is whether our organizations are learning as quickly as we are automating. 

 

Organizations that thrive will not necessarily be those with the most sophisticated algorithms. They will be those with the most sophisticated learning systems—organizations capable of detecting unintended consequences, questioning assumptions, and adapting before harm becomes embedded in the customer experience. 

 

The future of customer experience will not be shaped by artificial intelligence alone. It will be shaped by leaders who recognize that every decision deserves a learning loop, and remembering that behind every dataset is a human being whose experience still matters. Every system deserves leadership humility to ask, What are we missing before someone else has to tell us? 

 
 

References (APA 7th Edition) 

Argyris, C. (1977). Double-loop learning in organizations. Harvard Business Review, 55(5), 115–125. 

Cheong, M. (2024). Transparency and accountability in artificial intelligence systems: A review. AI and Ethics. Advance online publication. 

Marcineková, K. (2025). Human oversight and empathy in AI-enabled customer service: A systematic review. Journal of Service Management. Advance online publication. 

Meadows, D. H. (2008). Thinking in Systems: A Primer.  

NIST. (2023). AI Risk Management Framework (AI RMF 1.0).  

Rainy, T. A., & Goswami, D. (2025). Mechanisms by which AI-enabled CRM systems influence customer retention and overall business performance: A systematic literature review of empirical findings. Advances in Social Research Conference Proceedings. https://global.asrcconference.com/index.php/asrc/article/view/9 

Senge, P. M. (2006). The fifth discipline: The art and practice of the learning organization (Rev. ed.). Doubleday. 

Shneiderman, B. (2022). Human-Centered AI. 

Soulami, B., et al. (2024). Artificial intelligence in the workplace: Implications for employee well-being, engagement, and organizational performance. Frontiers in Psychology, 15. 

Tang, P. M., et al. (2023). Artificial intelligence and employee social affiliation needs. Journal of Applied Psychology, 108(9), 1457–1474.