Initial lessons from deploying AI
As we read daily, companies and governments are pouring huge amounts of money into AI transformation. Our management consulting friends at BCG estimate GenAI budgets alone are expected to grow 60% over the next three years, to account for c.10% of organizational IT budgets. McKinsey estimate only 1% of executives describe their AI rollouts as “mature” (BTW, I`m sure McKinsey are profiting handsomely on advising what “mature” looks like)
From my own recent experience rolling out a suite of new AI tools for my organization, it isn't really about the technology at all. It's about human behavior and the psychology of change. I`ve spent the last several months working on an enterprise roll out, and it seems to come down to some basic fundamentals: people need to feel safe experimenting; they need to feel competent (capable of using the technology), autonomous (choosing how to apply it to their work), and connected (part of a group that values this capability). So a lot of basic Psychologcal Safety, Self Determination Theory and Social Identity Theory seems useful here.
I recommend a 3 step initial framework
Step 1: Get senior managers & leaders hands-on with AI. Get people onboarded to your solution then give everyone a 5 minute challenge: “Upload your latest report. You have 2 minutes using AI to extract this data and create a summary that you would share with your line manager. Go.” This builds psychological safety. When people see their senior managers struggling, experimenting, and sometimes failing with AI, it signals that it’s safe to try and learn.
Step 2: Build prompting skills through hands-on workshops. Run 30 minute training sessions. Let people experience bad outputs from poor prompts first, then teach them a simple framework and watch them get dramatically better results. This builds competence,one of the three core needs in Self-Determination Theory (Deci & Ryan, 1985). The “desirable difficulty” principle also shows that experiencing failure followed by success creates deeper learning and intrinsic motivation,building the confidence people need to keep experimenting on their own.
Step 3: Use super users to spread the word. Identify natural champions. Work with them to discover high-value daily use cases, then coach them to demo for peers. This provides autonomy. People choose how to apply AI to their specific challenges rather than following rigid mandates. It also leverages social referents within group identity: when colleagues who share the same role, department, or challenges successfully adopt AI, they become powerful reference points for what “people like me” can achieve. Research shows that early adopters create evidence that others then feel expected to follow.
