We're All Talking About Artificial Intelligence, but Are We Listening?

Why we cannot market our way to AI adoption

You have just finished your mandatory AI e-learning. Now you are catching up on the company's latest AI strategy, and you will need to be quick, because there is an AI workshop to prepare for this afternoon. When you open your inbox, a newsletter from yours truly has just arrived. The subject? You guessed it.

But what happens when people are overloaded with one-directional communication about AI, and is all this talk actually working?

Research confirms that businesses that invest seriously in artificial intelligence are right to do so. When applied to the right use cases, AI technologies can add real value. Robotic Process Automation can take over repetitive, error-prone tasks and free people to do more complicated work. Anomaly detection in supply chains can catch problems which would otherwise go unnoticed for weeks. Fraud detection systems can identify suspicious transactions quickly and accurately.

Recognising the value of AI, many businesses are wondering how to build on that success and increase uptake of AI. One of the questions that businesses attempting to drive AI adoption are asking themselves is: how do we get people on board?

But a more fundamental question to ask is whether integrating AI technologies into a workflow solves an actual problem that employees or customers are facing. When AI removes a task that was tedious, or handles a volume of work no human could manage alone, adoption tends to follow naturally. In workflows where the value is unclear, adoption is slow and potentially even unnecessary. Most of us have encountered an example of this: an AI assistant layered onto a process that was already working fine, or a chatbot to answer questions that a quick search can resolve faster and more accurately. When Westenberger et al. interviewed AI experts about why AI projects fail, the critical factors were rarely the technology itself. Projects failed because of unrealistic expectations, misunderstandings about what AI can actually do, and use cases that never offered real value to begin with. When employees ask “why do I need this?”, that is not resistance. It is one of the many reasonable questions that deserve to be taken seriously.

Yet that is not always how these questions are met. In an attempt to drive user adoption, organisations may be tempted to use communication that flows almost entirely in one direction. Employees are told that AI will transform the way they work. Internal campaigns are launched, excitement is built, success stories are shared. This approach treats the usage of AI as a communication problem. As if the right message, delivered consistently enough, will eventually land. But is this approach working, and if not, why not?

The result of top-down communication is that adoption figures may look promising on a dashboard, but fail to deliver in practice. AI-assisted applications might be installed on all employee devices, but not actively used. And users who don’t truly understand a tool might use it in a way that wasn’t intended. They will rely on outputs they cannot verify. They will struggle to justify their decisions when questioned. And when something goes wrong, trust collapses quickly, often taking the broader AI initiative with it.

When adoption rates disappoint, the response is often to try harder with the same approach: more communication, more workshops, more marketing. But we need to be careful when we describe employees as “not yet convinced”. It implies that their current position is by definition temporary, that they simply require more convincing, and that the right communication will eventually bring them around. We are not asking if their concerns are valid, we are only asking how to overcome them. And nothing makes people disengage faster than feeling unheard.

Sahay, who interviewed executives and employees at large companies about how input was gathered during change, distinguishes between listening that is genuine and listening that only looks like it. Active listening means asking for input only when we are genuinely open to acting on it. When we invite input but have already decided on the outcome, people notice. In the study, this kind of surface-level listening was linked to more stress, more resistance, and a growing reluctance to share honest feedback.

In a review spanning 79 studies across sixty years, Oreg et al. found that people respond more positively to changes they had a genuine hand in shaping. When people conclude that their input changes nothing, they stop offering it. Morrison has spent two decades studying when employees speak up at work and when they stay silent. When employees stop voicing their opinion, the organisation does not just lose goodwill, it loses information that it needs. Widespread silence has been linked to stress, dissatisfaction, and disengagement.

In today’s polarised world with a tendency to think in extremes, the AI debate can easily create the image of two opposing camps trying to convince each other. On one side we have the “AI enthusiasts”, sometimes dismissed as blind followers of the AI hype. On the other side we have the “AI sceptics”, who can be seen as blockers of innovation or resistant to change. Neither label is valid or particularly useful. What moves things forward is genuine curiosity and dialogue.

When we approach concerns around AI usage as useful information rather than obstacles to overcome, we might find that “sceptics” are often asking exactly the questions that responsible AI deployment requires. What value does this tool add? Why does the system make the decisions it does? What happens when it gets something wrong? Who is accountable? They are the right questions, and organisations that take them seriously tend to end up with more sustainable, more trusted AI implementations as a result.

The goal of asking critical questions and raising ethical concerns is not to slow down innovation. It is to make adoption durable. The businesses that will build lasting value from AI are the ones whose employees genuinely understand what they are using, apply it appropriately, and can stand behind the decisions they make with its assistance. That kind of adoption requires investment in understanding, not just exposure. It is how AI finds its way into the workflows where it belongs, in processes where it adds value, adopted by people who understand what it does and why it helps.

When feedback is treated as a misunderstanding to be corrected rather than a perspective to be considered, the message people receive is that the conversation was never really open. Once we stop broadcasting and start listening, the conversation moves from convincing people to understanding them. Importantly, it is how we can include people in a conversation that affects all of us.

---

Thank you!

AI Ethics Made Practical is moving to a bi-weekly schedule with new articles every other Wednesday. There’s a lot to come in the months ahead: we’ll be covering the big questions in AI Ethics and human-computer interaction, and discuss how these topics apply to all of us in daily life.

---

Thank you for reading AI Ethics Made Practical! Subscribe for fee to receive new posts and support my work.

Help to spread the message and share this article.

AI Ethics Made Practical is free of charge. But if you found this article helpful and would like to support my work, please consider buying me a coffee!

---

References

Morrison, E. W. (2023). Employee Voice and Silence: Taking Stock a Decade Later. Annual Review of Organizational Psychology and Organizational Behavior, 10(1), 79–107. https://doi.org/10.1146/annurev-orgpsych-120920-054654

Oreg, S., Vakola, M., & Armenakis, A. (2011). Change Recipients’ Reactions to Organizational Change: A 60-Year Review of Quantitative Studies. The Journal of Applied Behavioral Science, 47(4), 461–524. https://doi.org/10.1177/0021886310396550

Sahay, S. (2023). Organizational listening during organizational change: Perspectives of employees and executives. International Journal of Listening, 37(1), 12–25. https://doi.org/10.1080/10904018.2021.1941029

Westenberger, J., Schuler, K., & Schlegel, D. (2022). Failure of AI projects: Understanding the critical factors. Procedia Computer Science, 196, 69–76. https://doi.org/10.1016/j.procs.2021.11.074

---