L&D in the AI era: Augment or upskill?
As you push for greater human performance, you have a choice to make.
Want to write an interactive story? There’s an AI for that. Want to solve an Excel problem? There’s an AI for that too. Want to find an AI for whatever random task you’re facing? There’s a website for that. It’s called: TheresAnAIForThat.com.
All of these tools enhance our existing capabilities. But, in most cases, they don’t do it by helping us improve our skills. Instead, they provide augmentation so that we don’t need those skills anymore.
Our friend Nick Shackleton-Jones likens this to an exoskeleton. If you’ve seen the movie Aliens, Sigourney Weaver battles the alien queen with the help of a loader suit. This suit gives her protection and sufficient strength to make it a fair fight.
But the technology only improves her capability so long as she’s suited up. Once the suit goes flying out into space, Weaver is no more capable of fighting an alien hand-to-hand than she was at the start.
So it goes with AI augmentation, except where Weaver is fighting aliens, I’m crushing Excel.
In the past week, I’ve used ChatGPT to write the first draft of a policy document, compose interview questions, and create complex formulas.
But strip me of my AI augmentation and there’s no way I could take a spreadsheet listing classic works of literature and COUNTIF ‘Monte-Cristo’ appears. (Pause for applause [1]).
Is this a problem?
It might become one.
A team of researchers in China and Australia recently warned of ‘metacognitive laziness’ after finding that students who were assisted by ChatGPT performed better on assessments but showed no greater knowledge transfer than their peers:
‘In the lab setting… some learners would subsequently copy and paste content generated by ChatGPT (e,g., example sentences given by ChatGPT) to achieve high scores by catering the scoring rubric. Therefore, we argue that this ”out-performance” might be the result of ”AI-empowered learning skills” which optimise performance at the expense of developing genuine human skills.’
A recent meta-analysis found similar, highlighting ChatGPT’s positive impact on academic performance.
But, as one of those original researchers wrote on LinkedIn, academic performance isn’t quite the same thing as learning or developing skills:
‘The issue with this meta-analysis is that it confuses learning with the product measures (e.g., essay score) of learning performance… This whole issue also highlights that GenAI has "hacked" our existing measurement and assessment approaches.’
This isn’t a new phenomenon. Humans have always used technology for ‘cognitive offloading’. The calculator made it easier for us to solve math problems. SatNavs and Google Maps make it easier to reach our destinations. But, in so doing, we lose our ability to do calculations in our head or remember how to get to the supermarket.
A recent (and pertinent) example from learning tech entrepreneur Ben Betts showed how an AI tool, Operator, can complete a variety of different e-learning courses for you. Without you even being there! You could go and make a coffee while your Data Protection course is completed.
And herein lies the problem: what was the intended purpose of that Data Protection course? To trigger completion or for someone to learn something? AI is hacking our measurement of compliance, just as it hacked our academic assessments.
For those of us working in learning and development, this should re-focus us on what we are trying to achieve. Not ‘tick box’ completions, but real improvement in the performance of the people we support.
To do this, we need to embrace both of the opportunites AI presents here: augmentation and skill development.
Or, what Marc Steven Ramos recently described as ‘the proportional value of… quantifiable tasks with the complementary value of… qualified skills’.
🚀 AI tools like ChatGPT can improve performance by helping us perform tasks more quickly and to a higher standard.
💪🏽 They can also, when used in the right way, help us perform more effectively by building our skills.
Evidence for this comes from The World Economic Forum’s recent ‘Future of Work’ report, which I wrote about a couple of weeks ago. The top skills sought by employers around the world are around technical literacy and AI (the ability to augment others); as well as leadership, creativity and agility (the ability to work well with others).
🧐 An example
To take a classic management example: imagine a manager needs to have a difficult conversation with a direct report.
First, they have a task to complete: prepping for that conversation. Our new AI coaching tool can help them do this by prompting them to think about the issue more deeply and by surfacing support resources. It’s part of our Content Hub.
Then, they have a skill to demonstrate: speaking to the direct report. The manager cannot stop midway through the conversation to ask an AI for advice (very awkward!). But they could use our AI Conversations tool to practice the conversation in advance and build their proficiency.
When you’re next faced with a workplace performance issue, here are three questions I think it’s worth considering:
What are we trying to achieve? Augmentation or skill development? Both are valuable, but distinct.
Are AI tools helping us create better outcomes, or just faster ones? We should think critically about our own usage.
Where we are creating digital learning experiences, why would anyone bother to complete them without an AI? If you can answer this question, you must be taking a more user-centric approach that is aligned with the learner’s needs and motivations. Well done, you!
If you’d like to explore how Mindtools can help managers diagnose, accelerate and evolve in their careers, get in touch! Email custom@mindtools.com or reply to this newsletter from your inbox.
[1] I ran this joke past my wife and she didn’t think it was funny. So I asked ChatGPT to explain why it’s hilarious.
🎧 On the podcast
Those of us who work in learning and development like to think we're in the business of behavior change. But we often don't have an in-depth understanding of what current behaviors are, or how to change them.
In this week’s episode of The Mindtools L&D Podcast, I was joined by pals Ross Dickie and Dr Anna Barnett to discuss:
methods for understanding behavior;
digital learning methods for changing behavior;
methods for measuring change.
Check out the episode below. 👇
You can subscribe to the podcast on iTunes, Spotify or the podcast page of our website. Want to share your thoughts? Get in touch @RossDickieMT, @RossGarnerMT or #MindToolsPodcast
📖 Deep dive
Last week, my partner-in-crime Ross Dickie wrote about Duolingo’s use of gamification to drive engagement.
In his piece, he quoted Duolingo CEO Luis von Ahn saying:
‘Always, primarily, we are a motivation engine.’
What’s interesting here is that von Ahn is clear on the purpose of gamification: It is to motivate app users to spend more time in the app.
🤔 Why would you include gamification in your learning course or program?
A 2021 meta-analysis from researchers in Spain identified various constructs, separated by educational level, that you could target:
Across almost all of the studies, gamification led to positive results for each.
The question for learning designers then is, what are you targeting? Motivation appears frequently, but how you gamify an experience to encourage academic achievement or participation is likely different to how you would for motivation.
Manzano-León, A., Camacho-Lazarraga, P., Guerrero, M. A., Guerrero-Puerta, L., Aguilar-Parra, J. M., Trigueros, R., & Alias, A. (2021). Between level up and game over: A systematic literature review of gamification in education. Sustainability, 13(4), 2247.
👹 Missing links
🎮 Wait, don’t gamify that thing!
In contrast to the ‘Deep Dive’ above, I thought I’d share this timely warning from Robert Meza. Just because you can gamify something, doesn’t mean you should. Robert points out that ‘engagement’ is your product’s first layer, but that this should be followed by layer 2: ‘The behavior you want someone to achieve outside your product’. With Duolingo, layer 2 could be ‘speaking French’. Often though, we never get to layer 2, because we just want people to get addicted to tapping.
🤖 Google’s new AI model is trained to be a learning designer?
The AI news keeps coming, and I know some of you hate it. But in this video, our pal Ross Stevenson demoes the capabilities of Google’s new LearnLM. According to Google, it inspires active learning, manages cognitive load, adapts to the learner, stimulates curiosity and deepens metacognition. Does it work? Check out Ross’s video to find out.
🧠 An agonizingly in-depth insight into DeepSeek
If you have much, much, more than a passing interest in the news that Chinese firm DeepSeek upturned the AI industry over the past couple of weeks, an accessible overview of what happened can be found in Trung Phan’s newsletter. Accessible, but unbelievably long. Fortunately, Trung uses memes to keep it light.
👋 And finally…
How close are we to completely replacing voice actors with AI? In this short documentary from The New York Times, we get our answer!
Hank Azaria, who voices Moe, Chief Wiggum and others on The Simpsons, puts an AI voice generator to the test.
👍 Thanks!
Thanks for reading The L&D Dispatch from Mind Tools! If you’d like to speak to us, work with us, or make a suggestion, you can email custom@mindtools.com.
Or just hit reply to this email!
Hey here’s a thing! If you’ve reached all the way to the end of this newsletter, then you must really love it!
Why not share that love by hitting the button below, or just forward it to a friend?