Why do you think that some devs are able to significantly improve their productivity using AI, while others aren’t?

High-performing developers use AI as a force multiplier because they think clearly and iterate, others struggle because they expect answers without providing direction.

The use of AI by developers to improve productivity depends on what they are trying to achieve. In my experience, pre-development activities have shown great success (e.g., seed code generation has saved our developers on average 20% of their time), and while in the midst of the development, AI toolsets have shown significant efficiencies not only through automation but also by reducing overall developer error, saving thousands of hours. In one such case study, we have augmented the DevOps process to save us 20-35% of the overall effort.
https://fulcrumdigital.com/blogs/building-smarter-systems-ai-in-devops-for-scalable-sre-operations/

Modernization is another area where the AI-led discovery phase has shown tremendous efficiencies. The attached case study has a live example of how we use AI in modernization projects.
https://fulcrumdigital.com/whitepapers/ai-dev-assist-unlocking-efficiency-and-modernization/

Two things come to mind: applicability and passion. For some devs it’s just not relevant - COBOL IDEs aren’t ready to receive value from AI.

In terms of passion, between GITHUB copilot, OpenAI codex, Anthropic, Perplexity and then a variation of DEX tools with private LLMs and now a slew of sourcing suppliers bringing in their own tools - it takes a passionate developer engaged nights and weekends to ver all of that on productivity..

Sometimes, simply saying there's value in using AI isn't enough of a motivator.
If you have a small enough group or specific individuals you want to motivate to use AI more, talk to them about it.

If you have a larger group, you can still get some feedback by sampling from the group, then prepare a talk on how the use of AI tooling is delivering productivity, specifying where, how, and why. You may have some individuals who won't accept a change in tooling, but most resistance is just due to people getting stuck in a flow.

I'd also listen to any strong detractors, as they may show that others' productivity isn't delivering the broad value it may seem to on the surface. i.e., Faster coding may be causing bugs or maintenance issues later.