From Hype to Human: What Graeme Codrington’s Talk Means for Leaders Who Care

06/10/2025

When technology gets loud, it's easy to forget the quiet work of being human. Graeme Codrington's recent talk at the MANCOSA Jacaranda FM Business Breakfast cut through the hype with a simple, liberating idea: use AI where it's strong, and double-down on the work only people can do—empathy, discernment, context, care. At MINDSHIFTERS, that resonates deeply with our coaching philosophy: transformation starts with people, not with tools.

1) AI is helpful—but it isn't wise

Graeme reminds us that today's models are probabilistic pattern-matchers, not thinking beings. They "guess" based on the data they've seen, which itself is biased and incomplete. That's why seemingly simple prompts (like asking for an image of a left-handed writer) so often flip to right-handed outputs; the internet's training images skew that way. The point isn't the party trick—it's the structural bias underneath our tools.

In South Africa, we also see how language, accent, and idiom confound machines. Human nuance—humour, context, code-switching—is where people shine and machines stumble.

MINDSHIFTERS takeaway: AI can draft, summarise, and translate—but it cannot hold meaning, values, or moral complexity for you. You still need wisdom.

2) Call the "manager": where human authority and care matter

Graeme's restaurant metaphor is brilliant: "Please may I speak to the manager" is where scripts and rules end and judgement begins. That's the line AI can't cross—exception handling, context, and compassionate authority. Your teams need leaders, not just tools.

MINDSHIFTERS takeaway: Train AI for routine execution. Train people for exception management, reconciliation, and relational repair.

3) Stop calling it "Artificial Intelligence"—think "Intelligent Assistance"

Treat AI as IA—intelligent assistance—that augments your work rather than replaces it. It's superb when it amplifies human strengths; it's dangerous when we abdicate human responsibility to it.

MINDSHIFTERS takeaway: Ask in every workflow: What should machines do here? What must humans own? Then design for dignity, not just efficiency.

4) Design a "Bionic" organisation (people-first, tech-powered)

Graeme's vision of becoming "bionic"—staying human with machine parts—is both practical and hopeful. Use technology to extend capacity, clarity, and consistency; keep humans at the centre for culture, courage, and care.

MINDSHIFTERS takeaway: Pair automation with coaching. Every rollout of a tool should be matched with a people plan: change stories, listening forums, psychological safety, and values-aligned guardrails.

A People-Centred AI Playbook (for Leaders & Teams)

1) Clarify the "Human Only" zones

  • Exception handling, conflict resolution, performance conversations, values decisions, and customer recovery. These are never fully automatable—own them.

2) Use AI for "Hard-for-Humans" tasks

  • Synthesis, pattern scanning, first-drafts, translation, tailored message variants for diverse audiences—then human edit for tone and context.

3) Build an internal bias check

  • Before deploying prompts/workflows, ask: Whose data shaped this answer? Who's excluded? (Remember the left-handed and language examples.) Add a human review step for equity and accessibility.

4) Coach for courageous conversations

  • Use AI to rehearse difficult dialogues, but never to replace them. Practise with role-plays; then meet in person to listen, adapt, and repair. (Machines can simulate; leaders must care.)

    5) Define your "manager moment" protocol

  • When a process breaks, who steps in, how fast, with what authority? Make this explicit so customers and staff experience humanity—swiftly.

6) Language with love

  • Let AI produce variants for different audiences; your leaders choose the words that build trust across cultures and generations.

Reflection Questions for Your Next Team Meeting

  1. Where in our customer journey do "manager moments" happen most—and how do we ensure human authority and kindness show up there?

  2. Which parts of our work are genuinely better, faster, and safer with AI—and which require human discernment?

  3. How will we mitigate bias in our prompts, datasets, and outputs before they impact people? (Think accents, idioms, representation.)

  4. What are our non-negotiable people practices (listening, debriefs, coaching) that keep us truly human-centred?

Our MINDSHIFTERS Promise

We help leaders build cultures where people flourish and performance follows. We'll bring the best of behavioural science and values-driven coaching—then help you use AI as a servant to that mission, not its master. As Graeme frames it: let the tech make us superhuman in execution, while we stay profoundly human in what matters most.

If you'd like support designing your "bionic" leadership system—policies, prompts, playbooks, and people practices—we're ready to walk the journey with you.

Share