Learning & Development: the new #1 organizational function

Organizations can't adapt if their people don't. And even if genAI is "done", we'll have to learn to do our jobs differently. That means the CEO's new priority should be L&D, and it needs to step up.

Welcome, new subscribers! Announcing a new book clearly got your attention. And a warm hello to those who were already here. Grateful for your continued enthusiasm!

I wrote the book to help a broad audience save human ability in this age of intelligent machines. Early feedback from people I trust and respect tells me I did my job right: it offers insight and practical help to anyone who picks it up - regardless of skill level, career stage, occupation, or role.

But for all our benefit, today I’m going to focus on one group close to my heart, part of organizations around the world: learning & development.

“Try again to show “The L&D genAI challenge”, but with very different art styles. Perhaps a stained glass window, an ancient grecian mosaic, a comic book style image, and a watercolor painting. Be very creative!

The professionals in this function curate the human ability that backs up their organizations’ promises to the world. Each of us owns our skills journey, of course, but L&D professionals are experts at what it takes to build relevant, valuable expertise in their context. Accounting, Marketing, Finance, Maintenance, Facilities, Sales, Recruiting, R&D - all these functions do work that keeps an organization running. L&D is special in that it does the work that helps all the other functions stay relevant in changing times. A poor L&D function means all other functions get dull through use. A good one helps all those saws keep the edge they need to handle the next challenge.

If you’ve been through a training, read an FAQ or a job aid, been through a job rotation, been a formal mentor or mentee, or had a performance review, you’ve relied on the systems and processes these experts put in place. When they do their work well, your skills journey is clearer, more motivating, more valuable to those you serve, and helps your organization adapt more effectively. I should know - I spent over a decade in this industry before I went back for my PhD, working at this problem from the outside as a consultant and from the inside as a manager. Professionals in this function are fascinated with how people learn best, and have the skill to make learning mean business.

So when I started to write The Skill Code about three years ago, I believed L&D professionals were key to their organizations’ adaptation to intelligent machines.

I just didn’t realize they would become mission critical.

Like most of us, three years ago I didn’t anticipate the generativeAI tsunami. Yes, I believed intelligent technologies would play a rapidly expanding role in the world of work. My wife and I bet my career on that belief in 2009.

But very few of us had the luck or the foresight to anticipate that we’d see the sudden arrival of a general-purpose technology that was available to billions of us for free - one that could perform a wide array of cognitive tasks very quickly and that interacted via an instantly familiar chat-based interface. Let alone that it would soon handle images, video, and audio, too. Let’s face it: though it’s weird, makes stuff up, and responds to social cues, to a naive user this technology is a shock. It seems intelligent.

The revelation - which I stand by - is now pretty clear: many of us will change the way we do our jobs to accommodate this new cognitive prosthesis. The now well-known “GPTs are GPTs” paper suggests that 80% of us could use generative AI to improve our productivity on 10% of our tasks. Many of us will go for it - staying put will often be like choosing to send faxes after email hit the scene. That’s an appreciable chunk of reskilling for 2.6 billion people. And if we take the combination of genAI and robotics into account, all of those figures are going to expand.

Without bold, immediate help from L&D this will hobble organizations.

Why? Without guidance, some employees will race *far* ahead with generative AI, but their collective sprint will be incoherent and uneven. Some will run in circles. Some will cause harm, or slow others down. Some will make stupendously valuable discoveries. And it will all be obscured in the welter of everyday operations: sales, customer service, perhaps even R&D might improve… a bit. Or key outcomes might just say the same. This is exactly what happened in my field study of robotic surgical training across 18 of the top teaching hospitals in the U.S.. Prior training methods remained in place, and got in the way. If you followed that playbook, you struggled to learn. A few residents found alternative ways to build skill anyway, and they raced ahead. But they hyperspecialized in robotic surgery, and their expanded learning opportunity came at the expense of the others in their learning cohort. Yet all graduates were legally empowered to use the tool when they left, and many did. It’s just that - in the words of one Chair of Urology - most of them “sucked”. So it’s no surprise that from a statistical point of view, robotic surgery was just as effective as prior methods. Analysts couldn’t take skill difference into account.

Things will be dramatically intensified here with generative AI, because unlike a thousand-pound, four-armed surgical robot, you can quickly use genAI for a wide variety of tasks, in a wide variety of ways, and it’s very easy to conceal the fact that you’re using it. People are going to scatter to the four winds with this stuff, and it will be very hard to learn from their successes and failures, let alone focus them.

Of course many, many folks will simply not do much at all, because they don’t know much about the new technology and how they might apply it in their day to day. And Microsoft’s bold play aside, it really hasn’t been integrated into familiar workflows in ways that take advantage of its rapidly expanding capabilities. At best maybe they’ll try something trivial with it, get a moderately interesting result, and settle for that. This variability includes managers, too: even if they got perfect, shared information on the potential upsides and downsides of implementing intelligent technologies (fat chance), they’d take different approaches to redesigning jobs and reskilling workers. The research on disruptive innovation and resource allocation makes it clear: some will implement bold, focused skill development experiments. Most will do a little something. Others will ignore the reskilling and work redesign challenge entirely.

Without top-notch L&D, introducing genAI into work is a bit like yelling fire in a crowded theater: it will tear organizations, teams, and careers apart from the inside. And I shouldn’t say will, actually. This isn’t future tense. It’s happening.

Leaders need to get wise to this, immediately. It doesn’t matter what industry they’re in. Their firm’s strategy is irrelevant. Whether they prioritize their sales teams, R&D function, manufacturing, supply chain team, service delivery, or any other group. They need to make bold, targeted investments in L&D to help their mission-critical functions adapt fast enough to capitalize on new opportunities and cope with new threats.

There’s a dangerous twist, though: the L&D playbook is out of date - follow it and you may well destroy more skill than you build.

Over the last 13 years I’ve found that the skill development game itself has been subtly disrupted from within - via the very intelligent technologies that are at the heart of the opportunity. The short story? The unifying feature of these technologies is that they allow us to self-serve far more than before. So we get productivity from them at the expense of collaborative interaction - most notably between experts and novices. The vast bulk of skill development occurs informally, right in the flow of work, so this is a profound threat, right when we need new skill the most. To add insult to injury, the traditional L&D investment is overwhelmingly weighted on formal solutions (e.g., training, job aids, mentorship) set apart from actual work performance. These need revision and professional attention, of course, but especially given the disruption to natural collaborative patterns, this emphasis is now officially backwards. My 2019 Harvard Business Review article sums up the argument, and I believe has stood the test of time.

All of this adds up to one, clear imperative: CEOs need to quadruple down on L&D as their new, #1 most critical business partner, and hold them to a higher standard than ever.

For the rest of this piece I’m going to get practical, and speak directly to the L&D professionals out there. If you want to show some leadership and renewed relevance in this crucible moment, I recommend the following:

Step 0: Put on your own genAI mask before helping the business

“I want an image in comic book style that somehow shows an L&D professional putting on a generative AI mask before helping other people in their business. Be fun and playful with this.”

As a matter of professional curiosity, you’ve probably at least tried out ChatGPT. And if you’re browsing around to stay current on AI, you may well have landed on material like Sal Khan’s impressive and inspirational Ted talk which included a demonstration of Khanmigo - an AI-based tutor for kids that has learning theory baked into it. It won’t give you the answer. It’s designed to teach you to fish. A wondrous vision for a complementary human + AI skills future, with proof of the first few steps.

Good for you. Staying one step ahead of those you serve is absolutely necessary to serve as a confident specialist partner in key organization design and management decisions. If you haven’t done this yet, you must do so, stat.

But you have to go deeper. Significantly so. To “get” AI and become a trusted, mission-critical partner for the business, you have to try the impossible: do some real technical work in one of the areas you serve, alone, with only generativeAI at your side.

Seriously. I wrote a post about this recently, and I believe you should treat it as a mandatory. The power, peril, and opportunity with generative AI is partly in how we can use it to do certain tasks incrementally better, but it is also in how we can use it to do things we never could before. Something we wouldn’t have even considered trying. In my post I show - in detail - how I challenged a class of Masters students to build a professional-grade web app for project management. The trick was they didn’t know how to code when we got started. That stopped them from pursuing lots of opportunities, and kept them out of the consideration set for technical project management in particular. Can’t build software? You’re not getting the job.

Let’s face it. You have long had a comparable problem: internal customers often see L&D (and HR) as a “soft” function with limited technical and quant fluency. You don’t “get” them, so from their point of view you can’t really serve them.

You have a golden opportunity to kill both birds with one stone here, just as my Master’s students did: get your own visceral sense of genAI while building credibility with those you seek to help. Only then will you be in a position to help your internal clients make wise decisions about how to redesign work in their areas, who needs what skill to support that change, and how to get it to them. You need to understand how genAI can be an ally - not a hindrance - in all this, too.

Here’s how you can adapt the full assignment I gave my Master’s students to achieve these goals. The first part is a solo act - a mirror to Phase 0 in my assignment. Don’t ask for any help, don’t google anything. Just try - and judge yourself by the criteria I list in that assignment: candor, grit, and creativity:

1. Choose a business unit focus, and design a synthetic dataset to analyze. Decide on a few of your clients’ key metrics (e.g., Sales: leads, conversion rates, revenue per customer, product performance. Software engineering: bugs, deployment frequency, velocity, backlog, dev satisfaction). Then decide on data types (numerical, categorical, text-based). Your choices should outline some part of the data that their function treats as mission-critical. Key to note: only one person in your function needs to create and validate the dataset and analysis tasks. If you’re first, congratulations! Share it - and your process learnings - with others taking this challenge. The point is to get to analyze (starting at step 5).

2. Create data. Since you shouldn’t share your company’s real data, program genAI (Gemini Advanced or ChatGPT 4.0 or Bing in creative mode) to create synthetic data for you to analyze. Yes, programming proceeds in natural language now:

   “Generate a synthetic dataset with 500 rows for a mid-sized software company's [sales] department. Include fields for [list specific fields you decided on], and ensure the data reflects [seasonal sales fluctuations] with a slight upward trend over a two-year period. I’m asking for this because I am an L&D professional and I want to try to do the kind of analysis that my [sales-based] clients do. Be sure to make this data realistic, including successes and failures, and at least one disruption, such as a competitor launch or a sudden regulatory change.”

3. Iterate with genAI to get this right. The initial dataset will have problems, and you’ll have to hang in there to correct them.

4. Choose goals. Use genAI to reformulate the phase 0 goals in my actual assignment to match the synthetic data and the real reporting needs of the business area. Give it the data, the goals I supplied to my students, and ask for a comparable set of goals that involve analyzing the data. Be sure they include descriptive statistics and basic plotting.

5. Analyze, using code. If you don’t already know python, create python code that completes these goals. If you python, use a different language you don’t know.

6. Debrief and learn. You know how to do this better than almost anyone in your organization: get together with colleagues after you’ve tried the impossible. Share chat transcripts. Share stories. Share successes. Failures. Concerns. Hopes. Questions. Consider the implications. Plan.

Now, at a whole new level, you’re credible with genAI, data analytics, python, and - perhaps most importantly - your internal customers’ business. You’re probably pretty far ahead of most of them, at least on some of this. Good.

A universe of possibilities and imperatives is now open to you. You can use genAI for rapid generation of customized learning material at scale. Or create “intelligent” knowledge bases that learn through interaction with users and phase out static FAQs and job aids. The list goes on.

Step 1: Focus on the threat to informal skill development

“Cover the idea of the next section on the threat to informal skill development with four images, in comic book style. I like the idea of a detective. Do two more with that, make them female.”

But for now, I’m going to pivot to the silent skills killer that lurks in your organization, right now. This is the one I alluded to above - one that’s going to become dramatically more dangerous as your internal clients get their hands on productivity-enhancing generative AI and start implementing it around their organization.

Let’s say they’re motivated to explore genAI possibilities. So ideally they get a fresh set of information about what their workforce does. What their tasks are. What those cost, and what the ROI is. Maybe you’ve even helped them with this analysis. Then they select an area where they believe their workforce and KPIs could get a great boost by integrating generative AI into the work. And they go for it.

If you stayed in pre-2023 L&D mode, you’d mostly help them with the reskilling plan to facilitate this. Who needs what kinds of training, job aids, and learning infrastructure to set them up for success? How will you know when you’ve gotten there? You might also help a bit further upstream in job design. Then you’d implement to support, lather, rinse, repeat.

Here’s the trouble. They - and you, if you’re not careful - will be sacrificing on the job learning on the altar of that short-run productivity. The immediate first crop of users will - fates willing - get better results for themselves and the business, but they will do so in a more self-serve way that limits vicarious learning for other people in the firm. Junior members of their occupation. Nearby administrative assistants. Colleagues. Even folks more senior to them. Less participation in real work means less challenge, exposure to complexity, and human connection. I explore each in detail in my upcoming book, but they are the lifeblood of the kind of informal learning we take for granted - and that has underwritten our progress for thousands of years. This self-serve threat to the expert-novice connection was real before intelligent technologies, but it’s even more true now.

To step away from more obviously intelligent technologies - and perhaps closer to home for you - consider the work from home trend that emerged during the pandemic. People who know my work regularly message me, worried about their adult children’s skill development at work, for example. Here’s a typical one from Ezra Zuckerman, a Deputy Dean at the Sloan School of Management at MIT, publicly on twitter (notably *after* Covid):

Work from home was a safety issue at the height of the pandemic. Then it became a productivity opportunity. And when surveys show morale improvements with remote work, it’s hard to pin down reasons to bring people back together. Skill development is a critical one, but its signs are subtle, and the cost of inaction shows up years later in ways that are very hard to pin to their original causes.

L&D’s two moves: Skill + AI and Skill x AI

“Now end with something more figurative, but in comic book style - to demonstrate more metaphorically the ideas in the last section. Here's the text:”

L&D needs to flip the script. It’s time to invest much more heavily in the informal learning to make up for the gaps in this age-old learning infrastructure. I detail some of the implications of this in my Harvard Business Review piece above, and go into much greater detail in the book.

Sometimes this will mean a “Skill + AI” scenario: L&D professionals can find ways to design, implement, or maintain technologies so that productivity and human ability are enhanced. The basic question here is “how can we get value from this technology and ensure that users build valuable skill as they deal with it?” A tall order. Not always possible. But we’re often not even trying. CEOs and managers, this is your chance to make your L&D folks your strategic partner. Considering a big new investment in technology? Bring your L&D leader to the table and let them ask this question of each vendor. Back them up. Ask vendors to do better. And when it’s time to implement, keep the question alive - for managers, for workers, and for your leadership.

But the deeper play is yours, too: Skill x AI. Your organization can race ahead - and the quality of work life can significantly improve - if you work towards a genAI-enabled network of human experts, novices, and AI, focused on building human and AI capability right in the middle of work. Many will push for a “one AI size fits all work” approach. You know better. My Step 0 assignment and your prior expertise put you in a fantastic position to assess which processes could be amplified or extended with these tools. This goes way beyond AI-assisted matches between experts and novices who happen not to work in the same physical space or organization. This is literally a new fabric for the expert-novice connection—where simply by engaging with it, both humans and AI learn faster than they could on their own, enhancing human relationships and our sense of fulfillment along the way. A lot of the tools to enable this are already on the table - in fact we have decades of pre-genAI research that shows automation can play a positive role in facilitating skill development.

It’s time to put them all to work in the smartest way possible to help our organizations adapt quickly and sustainably. The L&D function was made for this moment, and can be remembered as the internal leaders who rewrote the skill playbook just in time to avoid the skills iceberg above and capitalize on the incredible opportunities ahead.

So go ahead. Give yourself that Step 0 challenge. Connect with each other, and your leadership. And get to work. We need you more than ever.