4. Environmental impact
In the last email (Subject: Germ Theory), I made the claim that for L&D to stay relevant, we need to prioritise designing the environment in which people operate ahead of the capabilities of the individual.
Whilst skills matter, the greatest opportunities for performance improvement usually come from improving the environment in which those skills are applied.
Why?
Because when performance depends on people remembering, interpreting, improvising and compensating, it becomes less reliable.
Systems reduce the need for heroics. They make good performance easier, more consistent and less dependent on individual effort.
And the best performance partners have always understood this. They don't just help people perform better - they make work easier. Without a well-designed system, results are largely dependent on individual heroics.
But unlike people, AI can’t compensate for ambiguity, inconsistency and missing context, which makes relying on heroic effort an increasingly fragile strategy.
From the earlier email ‘Learning & Development 2.0’:
“For AI to perform to the required standard, it needs clarity. Which means for anyone using AI to perform to the required standard, they must provide AI with that clarity. And unless that clarity has been defined ahead of time, they'll be forced to figure it out for themselves, at the moment they undertake the task.”
Now, as I mentioned in that email, I want you to interrogate my claims with scepticism and an open mind - am I overstating the importance of the environment? Surely skills still matter?
Of course they do. But if both skills and environment influence performance, wouldn’t it make sense to start by designing the conditions within which people work before trying to improve the people? Otherwise, we’re putting the cart before the horse.
And if so, does that challenge some of our assumptions about the role of L&D?
Well, it’s an objection I hear regularly - we’re incredibly proud of ‘helping people’. For many of us, it’s why we do this work. But is it possible that by focusing on people and their skills, we're unintentionally limiting what those people can achieve?
To make the point, let’s use an example: ‘critical thinking’.
I've chosen that because I want to discuss a soft skill - something that traditionally, we might assume depends on the individual.
And it's also topical. With the advent of AI, there’s lots of bluster about the need for better judgement when it comes to evaluating outputs.
So, how would we improve someone’s ability to ‘think critically’?
Well, if we looked at this as a knowledge problem, the solution might be to deliver an awareness campaign, reminding people that AI can hallucinate and deliver substandard outputs.
And whilst this is better than nothing, knowing one should do something doesn’t mean one's behaviour will change (I know birthday cake is bad for me but it’s tough to say no when you’re only halfway through a fairy princess party and need a sugar hit before musical chairs).
If we looked at this as a skill problem, the solution might be to provide practice activities: “here’s an AI output, now evaluate it to ensure it’s appropriate, and we’ll give you feedback on what you missed”.
On the face of it, that doesn’t seem like a bad idea.
But dig a little deeper and we spot some problems…
Firstly, how is someone supposed to "think critically" in relation to that specific AI output? What are they looking for? How do they know if it’s acceptable? Against what criteria are they assessing it?
Secondly, even if they did know how to evaluate it, will they remember that in six months' time? Or will they still do it when it’s 5pm and they’re late to collect their toddler from nursery?
And what about Trevor who was wiped out with man flu when we ran the workshop - how do we ensure he thinks as critically as those who've done our practise activities?
So, whilst practise is important, I'd argue the first step isn't building the skill - it's defining what good critical thinking looks like in that context. And then baking that into the flow of work.
How might that look?
Maybe:
examples of good and bad outputs
a short checklist that appears with the AI output: “before sharing, check against these five criteria”
peer review before sharing sensitive work
Now, once this checklist has been designed, of course it makes sense to give people some practise using it. Throw them some edge cases, force them to navigate tricky scenarios, build the muscle of using the checklist so using it becomes second nature.
See what we're doing here? Environment first, skills second.
What we're not doing is relying on people remembering how to 'think critically' - we're designing the work so it’s part of the process.
And so, if this approach is true of critical thinking, could it be true of other skills too? Skills we’ve historically considered “soft” and which therefore come down to individual capability?
If so, we have some thinking to do - because if performance improvement starts with the environment rather than the individual, what does that mean for those of us who've built our careers around helping people learn?
We'll tackle that next.
- Ant
