Why AI adoption is a people challenge
AI value depends on people changing how they work. This module explains why adoption stalls — and what leaders can do about it.
Learning objectives
- Explain why AI adoption is mainly a people challenge
- Identify common barriers and responses
- Describe the adoption curve and J-curve
- Outline the leader's role in AI adoption
1.1 Technology is the easy part
Organisations rarely fail at AI because the models don't work. They fail because people don't use the tools, use them badly, or use them for low-value tasks — and because processes, incentives and roles stay unchanged. Getting real value means changing how work is done, and that is a leadership and change challenge.
1.2 Why people hesitate
| Barrier | What you hear | What helps |
|---|---|---|
| Fear for jobs | "Is this here to replace me?" | Honest communication about intent; involvement in redesign; reskilling |
| Lack of skills | "I tried it once; it was useless." | Role-specific training with real tasks; champions |
| Lack of trust | "It makes things up." | Teach verification; start with low-risk tasks; show sources |
| Unclear rules | "Am I allowed to put client data in this?" | A clear, simple policy and approved tools |
| No time | "I'm too busy to learn something new." | Protected learning time; leaders modelling use |
| Unclear value | "What's in it for me?" | Use cases from their own work; peer stories |
1.3 The adoption curve
New tools spread unevenly. A few enthusiasts experiment immediately; a larger pragmatic majority waits for proof that it helps people like them; some remain sceptics for good reasons (quality, ethics, workload). Leaders should equip the enthusiasts to become guides, give the majority practical evidence, and listen to sceptics — their objections often reveal real risks.
1.4 Expect a productivity dip
When people change how they work, performance often dips before it improves — the so-called J-curve. They're learning, tools need configuring, and processes haven't caught up. Plan for it: set realistic expectations, protect time for learning, and don't judge success in the first few weeks.
1.5 The leader's role
- Set direction — why AI matters for your team, and what it's for
- Model use — use the tools visibly yourself and share what you learn, including failures
- Make it safe — permission to experiment within clear rules
- Remove obstacles — tools, access, time, training
- Redesign work — change processes and goals, not just add a tool
- Measure and celebrate — recognise people who find valuable uses
For your team, rate each barrier in 1.2 as low, medium or high. For the two highest, write one action you could take in the next month.
See a worked example
Example — finance team of 25: Fear for jobs: medium. Skills: high. Trust: high (accuracy matters). Rules: medium. Time: high at month-end. Actions: (1) two 60-minute hands-on sessions using our own reconciliation notes, scheduled mid-month; (2) a one-page 'how to check AI output for finance' guide with examples of errors.
Key takeaways
- Models working isn't enough; work must change.
- Fear, skills, trust, rules, time and unclear value hold people back.
- Equip enthusiasts, convince the majority, listen to sceptics.
- Expect a dip before gains.
- Leaders set direction, model use and remove obstacles.