[Prosci Expert Insight] AI in Hong Kong: Adoption Is Up. So Is the Pressure.
Updated: Aug 14
Why redesigning work — not more training — is the change management problem that releases the pressure.
Two recent Microsoft announcements point to a tension every Hong Kong executive planning an AI rollout needs to sit with: 75% of AI users in Hong Kong fear falling behind if they don't adapt quickly. Yet 57% say it feels safer to stick with their current goals than redesign how they work.
Same people. Both true at once. That's not indecision — that's pressure with nowhere to go.
The talent is already there: 18% of Hong Kong AI users are "Frontier Professionals," ahead of the 16% global average. Having more Frontiers working alone may not close the gap.
The gap is a shared way to turn that individual use into organizational value — and until it exists, the pressure has nowhere to go but up. More training won't build it. This piece explains why, and what will.
The Real Challenge: Is There a Way to Let the Pressure Out?
That shared value doesn't happen on its own — here's why.
Beneath the headline numbers, the Work Trend Index reveals something more subtle — and more important for leaders to understand.
Two paths, Both Carrying Weight:
Adopting AI within current goals feels safer — even without measurable value — amid the fear of being left behind.
Redesigning how work gets done feels uncertain, risky, unrewarded, and unsupported.
The organizational readiness this creates is, in Microsoft's words, "limiting impact and increasing pressure on employees."
Prosci found that human factors dominate AI implementation challenges, accounting for 56% to 64% of reported difficulties across organizational levels.
Old work patterns can hold back new capability.
In practice, we see this play out as a reinforcing loop:
More individual AI adoption without a shared purpose → more fragmented AI practices and outputs across departments → less clarity on how information and decisions flow between teams → more pressure on individuals to deliver despite that unclarity → more training thrown at the symptom rather than the cause → which doesn't touch the fragmentation, so the pressure builds again.
This is a pattern we observe repeatedly in practice, not a sequence the Work Trend Index itself measured — but it's consistent with what the data does show: human factors, not tooling, drive the majority of AI implementation difficulty.
---
The Deeper Shift: From Functional AI Use to Cross-Functional Value
AI is often adopted within individual functions: one team uses it for customer insights, another for reporting, another for knowledge management. But value rarely stays within a single function. Insights generated in one part of the organization affect decisions in another. Without shared standards, structured validation, and cross-functional collaboration, AI can create fragmentation — not clarity.
But redesign is hard precisely because people are not moving at the same pace: some are ready to rebuild how they work, others are still deciding if it's safe to try. Engage too late, or engage everyone the same way, and the reluctant majority disengages while the frontier minority races ahead in isolation. Left unaddressed, that gap doesn't stay neutral — it shows up as burnout, disengagement, or quiet resistance.
This is not a training gap. Training assumes people already have the confidence and time to learn a new skill. What's missing here is sequencing: knowing who to bring along first, what support each group needs, and how to keep the fast movers from outrunning everyone else. That's change management's job, not L&D's.
The real work of AI-enabled transformation is not only about maximizing individual AI usage. It is redesigning how work, decisions, data, validation, accountability and collaboration flow across functions — while bringing people through that redesign at a pace they can actually sustain.
---
What the Data Tells Us About the Path Forward
Microsoft's research points to what actually makes a difference. Organizational factors — culture, manager support, and talent practices — drive twice the AI impact of individual factors alone. And when we look at what separates Frontier Professionals from others, the pattern is clear:
Manager Behavior | Frontier Professionals | Others |
Manager sets clear quality standards for AI work | 79% | 59% |
Manager creates space for experimentation | 80% | 61% |
Manager encourages ambitious work redesign | 81% | 63% |
For adoption, leaders need to manage two objectives at the same time:
Reinforce and Channel Frontier AI users into cross-functional value creation — so Frontier Professionals do not become isolated pockets of experimentation, but help improve processes, decisions and outcomes for end users.
Move less mature and non-AI users through their own adoption journey — building Awareness, Desire, Knowledge, Ability and Reinforcement so they can participate confidently and remain relevant.
Without both, AI adoption can split the organization: advanced users accelerate in silos, while others feel left behind, increasing fragmentation, resistance and pressure.
This aligns closely with Prosci's view of people managers as essential change roles. In Prosci terms, managers are not just implementers of AI policy; they are the local enablers of individual ADKAR progress:
Clear AI direction from managers supports Awareness
Psychological safety and experimentation support Desire
Role-specific guidance supports Knowledge
Practice and coaching support Ability
Recognition and incentives support Reinforcement
Here's the answer to the question from the opening: will more comms and training close the gap? Look at what training alone actually covers — Knowledge and Ability. It teaches people how to use the tool. It does nothing for Awareness of why the change matters, Desire to actually change how they work, or Reinforcement that makes the new behavior stick. That's the gap most "AI adoption" budgets miss — and it's exactly why adoption can stall even after the training is done. Closing all five requires change management, not a training module.
These are not technology investments alone. New AI tools can enable the redesign — but only if paired with the leadership behaviors that make redesign safe to attempt, reduce pressure, and build readiness across the organization.
---
What This Means for Leaders Now: Forward-Deployed Engineering Is Becoming Applied Change Management
Microsoft's Frontier Company announcement reinforces this from a different angle. Microsoft is investing $2.5 billion in a model that combines AI engineering with change management and continuous improvement. The signal is clear: AI value is not a deployment outcome. It is a change outcome.
For Hong Kong organizations, this moment calls for a different kind of leadership conversation — not only "how do we get more people to use AI?" but:
- How do we redesign work so AI reduces pressure on individuals rather than adding to it?
- How do we build readiness through the design journey, not as a separate training program?
- How do we ensure AI-generated insights are structured, validated, and valuable across functions?
- How do we measure value in terms of what matters to the people we serve?
As our Chief Global Officer, Mark Dorsett, puts it:
"Forward-deployed engineering is becoming something broader than engineering. It is becoming applied change management. The best deployment teams will still need deep technical capability. But they will also need to build trust, align stakeholders, surface resistance, translate value, and help organizations sustain new ways of working after the first successful demo. That is not a soft skill set. It is a deployment skill set."
Last Thought
Go back to that tension we opened with: 75% of your people are afraid of falling behind. 57% think it's safer not to change anything. Both numbers are a symptom of the same thing — employees who are willing, but unsupported.
A new operating model doesn't resolve that tension by moving faster. It resolves it by giving people a structured, shared way to redesign work together — instead of each person deciding alone, under pressure, whether to take the risk.
This isn't a fully new challenge. Organizations have navigated cross-functional process change for decades. What's different now is the speed and scale at which AI generates new expectations — and how little time employees have to catch their breath between them.
For Hong Kong organizations, the next advantage won't come from AI access. Most of you already have that. It will come from whether you build the change capability to turn individual AI use into shared, sustained value — before the pressure finds its own release.
One question worth asking in your next leadership meeting:
We've trained people on AI. Have we defined what success looks like, measured it, and built the behaviors to sustain it — or just assumed adoption would get us there?
Source:
Author： Catherine Tam Thillainathan
"We guide individual CHANGE from data to impact." Passionate about data and science, Catherine connects practitioners to turn vision and insights into action. A Prosci Advanced Instructor, Principal Advisor of ChangeAccomplishment, Cofounder of Master Change Circle, a lifelong learner, and a volunteer, she fosters peer collaboration and stays grounded in what matters—driving real results for the community.
Comments