Step 1: Create a Sense of Urgency
Kotter’s first step is deceptively simple: before people will change, they must feel that change cannot wait.
For ethical AI, urgency is not difficult to establish: the evidence is abundant. Facial recognition systems with documented racial bias. Hiring algorithms that discriminate against women. Predictive policing tools that entrench historical inequities. Credit scoring models that penalise poverty. Healthcare algorithms that underserve Black patients.
The challenge is not finding examples. The challenge is making them feel real and relevant to people in your organisation who may believe: sincerely: that their AI systems are different, better, more carefully designed.
Creating urgency for ethical AI means bringing the consequences home. Not as abstract statistics, but as specific, human stories. Not as someone else’s problem, but as a risk that your organisation faces right now.
It also means connecting ethical AI to organisational self-interest: not cynically, but honestly. Regulators are paying attention. The EU AI Act is law. Reputational damage from AI failures is swift and severe. Talent increasingly cares about the ethics of the organisations they work for. Urgency built on genuine risk is more durable than urgency manufactured from fear alone.
Step 2: Build a Guiding Coalition
No individual, however senior or committed, can drive ethical AI transformation alone. Kotter’s second step is to assemble a guiding coalition: a group with the authority, expertise, credibility and commitment to lead the change.
For ethical AI, this coalition needs to be genuinely diverse. Technical experts who understand how AI systems work. Legal and compliance professionals who understand regulatory requirements. HR leaders who understand workforce implications. Communications professionals. Frontline staff who actually work with or are affected by AI. And critically: representatives of communities most likely to be impacted by AI decisions.
A guiding coalition composed entirely of technologists will produce technically sophisticated but ethically narrow solutions. One composed entirely of lawyers will produce compliance-oriented frameworks that miss the human complexity. Diversity of perspective is not a nice-to-have : it is a functional requirement for ethical AI governance.
The coalition also needs visible executive sponsorship. Without it, every obstacle becomes a blocker. With it, obstacles become problems to be solved.
Reflection question: Who in your organisation would need to be part of a guiding coalition for ethical AI? Who is currently missing from those conversations?