Before we can help people build workflow skills and leverage AI in their work, we need to understand our audience. In the third part of this series, I unpack the different mindsets people adopt, which often limit their progress, so we can design the right experiences for each.
This is the third part of a series on building workflow capability for enterprise AI, the previous two parts being:
Part 1: Why AI Will Fail Without Enterprise-wide Workflow Capability
Part 2: How To Design An Enterprise AI Workflow Capability Program
In this article, we'll look at the transformation people need to experience, because understanding our audience will help us design an effective solution.
The mindset required to leverage AI
Let’s begin with a hypothesis: people can only improve their own work with AI once they have the correct mindset.
Now, this might sound unecessarily complicated, but AI has thrown up all kinds of resistance within the workforce. For example, we’ve probably all heard variations of the following (or maybe even thought these things ourselves):
“AI will take my job".
“I don’t need to give AI clear instructions because it can fill in the gaps”.
“If I define my workflows, my company won’t need me”.
Now, it may be that some of these claims are true! We’re not here to debate that…
But what we are here for is to design conditions that minimise these concerns, so our people can adopt the mindset needed to master the skills needed to leverage AI in their work.
And after having taught thousands of people to improve their work with AI, I have seen many objections that stop people from achieving our desired outcome.
And they all fall within one of the following six phases, which are usually a progression:
Mindset 1: “I don't want AI to help with my work”
Most people are past this stage now - but this is the classic laggard who is anti-AI. This could be for a number of reasons - environmental, social, philosophical etc. We aren’t here to judge.
What we might hear:
"I don't trust it."
"It makes mistakes."
"I enjoy this part of my job."
What's really going on: This is rarely about the technology. It's usually a mix of fear, pride in work they do well, and a past experience that confirmed their doubts.
What shifts it: Not persuasion. Pushing someone to adopt AI tends to harden the position. Give them a low-stakes reason to look (a task they already dislike) and the reassurance that they stay in control of the output.
For L&D: Don't design the workshop around this person. Make it easy for them to wander in.
Mindset 2: “AI can figure out to do my work by itself”
B: I want AI to help with my repetitive, manual work. But I'm not convinced AI needs me to define my work before it can deliver consistent results.
What you'll hear: "Why would I need to define my work? AI can do that for me!"
What's really going on: These are the people I call the AI delusionals (alt: "Hopefuls"). These are the folks who believe AI can magically intuit everything that's in their heads.
Whilst this is better than those unwilling to even consider using AI, this person suffers from a misunderstanding as to how the technology works. AI is using information sourced from the LLM, and there is no way it could guess how you work.
Not only do they mistakenly believe AI can guess all the nuance involved in doing their work, they also fail to recognise that if it could do their job, they would have a significant problem in that it won't be long before AI will be replacing them!
What shifts it: A fair test. Have them give AI a one-line instruction for a piece of their real work, then compare the result with what they'd actually accept. The gap does the persuading. [Example from your workshops?]
For L&D: And so, our goal with this persona is to help them understand that this skill of defining their work is a very human skill. Yes, we can leverage AI to help us, but it's not something that we can outsource. Design a safe place to be disappointed by AI, then show that the missing ingredient is theirs to supply.
Mindset 3: “Sharing my expertise will reduce my value”
C: I know that for AI to help with my repetitive, manual work, I need to define what I'm doing and how I'm doing it. But I'm worried that sharing my expertise will lead to AI removing some of my current responsibilities and reducing my value.
What you'll hear: "Won't AI replace me once it knows what I do and how I do it?" "If I write down everything I know, what's left for me?"
What's really going on: This is a fair fear, not a mindset flaw, and dismissing it destroys trust.
What shifts it: Those manual, repetitive tasks will eventually be done by AI anyway - so why delay the inevitable? I'd much rather be at the bleeding edge of those changes i.e., the person implementing them. Because then I can be proactive about where I'm focusing future efforts (hint: those activities AI can't do). Add an honest conversation about what AI can't do (judgement, relationships, accountability), a decision about which steps stay human-only, and seeing that the person who can define the work is the one who decides what AI does.
For L&D: Let people choose what they define and what they keep. Safety comes from control. [Link to your later post on resistance to giving away expertise.]
Mindset 4: “I don’t need help defining my work”
D: I can see how defining repetitive, manual work will help me leverage AI and enable me to be more valuable, regardless of whether my role changes. But because I'm competent at this work, I'll naturally be good at defining what I do and how I do it.
What you'll hear: "I've done this for years, I could do it in my sleep."
What's really going on: These are the insightful people who recognise that to leverage AI effectively, we must define our work. But: many will assume that because they're good at their work, they should be good at defining it. However, this is a faulty assumption – and without prior experience or tools to help, all of the nuance stored in our subconscious will stay there.
What shifts it: And so, the mindset shift for these folks will largely come from the experience of defining a workflow – and realising it's more difficult than they expected. (A good analogy here is riding a bike – it looks so easy, until you climb on for the first time and realise you might need some help.) Experience, not explanation. Telling an expert they're overconfident won't land. Asking them to break down something they know well, to the level a new starter could follow, will. [Tea-making example from Part 2 or your phone example from Part 1.]
For L&D: The workshop's first job is to create a productive struggle that people can't argue their way out of.
Mindset 5: “I don’t need to practise defining my work”
E: I recognise that being competent at my repetitive, manual work doesn't necessarily mean I can effectively define what I do and how I do it to the standard AI requires. But I think I can do this well without using the tools or practising the skills.
What you'll hear: "Okay, it's harder than I thought. I'll just be more careful next time."
What's really going on: They've felt the gap, but they're treating it as a one-off slip rather than a skill.
What shifts it: Recognising what you wrote in Part 1: being able to do something is entirely different to being able to articulate how you're doing it. This is a skill - it's a skill that very few people have. And it's a skill that requires practice, guidance and support. A second attempt, this time with a simple rubric (see Part 4), makes this visible: when people define one workflow unaided and then another with support, the improvement is obvious and the case for tools makes itself.
For L&D: Don't hand over the tools at the start. Let people feel why they need them.
F: I recognise that using the tools and practising the skills will help me define my repetitive, manual work and accelerate the results I can achieve with AI.
What you'll hear: "Can I do this for my other workflows?" "Can my team use this?"
What's really going on: This person is ready. As you wrote in Part 1: people aren't lacking tech skills - they're lacking awareness of what's needed, an opportunity to practise and simple tools to prepare.
What shifts it: Nothing. Get out of the way, and give them resources they can pull from at the moment of need.
For L&D: Capture their stories. They become the experience that moves people at stages A to E.
What this means for L&D
And as L&D professionals, this is important.
We must:
deeply understand the mindset shifts needed to navigate this transformation,
master the skills needed to define repetitive, manual work, and
design an environment in which our colleagues can do the same.
I.e., we need to walk the walk before we can talk the talk.
Any organisation has people at all six stages at once. That's why generic training fails: one session has to reassure (A and C), challenge (B), surprise (D), equip (E) and release (F).
Over the next few posts, I'll share what this looks like in practice:
how to define a workflow in enough detail for a person or AI to execute it
how to design the workshop that gives people the chance to practise, and
how to build the self-service tool that guides them through it.
And if you're an L&D professional who wants to master workflow improvement and become the person responsible for unlocking the value of AI in your organisation, join my email list today.
Next article in the series: Part 4: