Step 7: Sustain Acceleration

One of Kotter’s most important insights is that change initiatives frequently declare victory too soon. The first impact assessment is completed, the first training programme is delivered, the first AI ethics committee meets: and leaders conclude that ethical AI is now embedded. The urgency fades. Resources are reallocated. Attention moves elsewhere.

And then, gradually, the old ways return.

Sustaining acceleration means using early wins as platforms for deeper change, not as endpoints. Each success creates an opportunity to ask:  what more is possible? Which remaining obstacles can we now address? What have we learned that should change our approach?

For ethical AI, sustaining acceleration might mean:
– Expanding the scope of impact assessments from new deployments to existing systems
– Moving from voluntary ethics reviews to mandatory ones
– Building ethical AI into procurement: requiring third-party AI suppliers to meet the same standards
– Deepening community engagement, moving from consultation to genuine co-design
– Sharing learnings publicly, contributing to the wider field of ethical AI practice

Sustained acceleration also requires continued investment in capability. Ethical AI is not a static destination. The technology changes. The regulatory landscape evolves. New ethical challenges emerge. The organisation needs to keep learning.

Step 8: Institute Change in the Culture

Kotter’s final step is the most profound:  change is not truly embedded until it lives in the culture. Until the new way of doing things is simply how things are done: unquestioned, unremarkable, normal.

For ethical AI, cultural institutionalisation means:

New leaders are selected partly for their ethical AI orientation: it is a criterion in hiring and promotion, not an afterthought.

Success stories are told and retold: the organisation’s narrative includes accounts of ethical AI in action, reinforcing what the culture values.

Onboarding includes ethical AI: every new employee, from technologist to marketer, learns about the organisation’s commitment to ethical AI and what it means for their work.

Ethical AI is connected to organisational identity: not “we do this because we have to” but “this is who we are.” The difference between compliance and conviction.

Cultural change is slow. It outlasts any individual initiative, any executive tenure, any regulatory requirement. It is the most difficult part of Kotter’s model and the most important. An organisation that has institutionalised ethical AI in its culture will navigate future AI challenges: ones we cannot yet anticipate: with far greater wisdom than one that has merely implemented a framework.

Reflection question:  What stories does your organisation currently tell about its relationship with AI? What stories would you want it to tell?