Take control! Don’t let AI define you!
If you didn’t have limitations and constraints (budget, technology, policy, culture, skills, etc.), would you work as a learning professional the same way you do today?That’s the ultimate question for AI-augmented learning jobs. We need to rethink the fundamental questions of why and how we do business. who are we? And how do we create value? The first obvious improvement with AI is always efficiency, reducing the amount of time spent creating content (measured in human equivalent time). That’s the beginning. Or is it a dead end?
In this article, let’s take a step back from the field of rapid content development using AI. Let’s take a look back at your current workflow. From the moment of training, through design, development, implementation, and perhaps measurement and evaluation.
In this article…
How did we get here?
Back then, there were separate groups building and delivering learning solutions. They were instructional designers (and sometimes instructional system designers), developers, and operations personnel. ID and ISD were responsible for working with the SME to create a storyboard for approval as well as proactively assessing needs. Once the storyboard was approved by everyone (final_final_gold_reallyLastVersion_useThis.doc), development began.
Developers took ideas from the paper and implemented them in Flash and other advanced tools. Changing something after development was painful and could cause significant delays. Another group, L&D operations, uploads this package to the LMS and tests it. If this was an instructor-led approach (in-person or virtual), we would have had to create guides, workbooks, slides, etc.
This was a very time-consuming and piecemeal process. Businesses began demanding faster reaction times. That’s where rapid e-learning tools come in. What Lectora, Storyline, and Captivate (to name a few) did to the eLearning industry was blur the lines between designers and developers. Now everyone can build things. And now anyone can design things. (Whether those “things” are effective or not is another matter.)
Those who became experts in both design and development had the advantage of being able to prototype, iterate, and improve themselves very quickly, creating a magical learning experience. They can quickly translate ideas from small businesses and stakeholders into dialogue.
However, until now, the basics of the workflow output remained largely unchanged. It’s just more efficient. Just like a warehouse looks good because it’s producing products efficiently.
From warehouse to general contractor: What AI really changes in L&D
Speaking of warehouses. Let’s explore an analogy to corporate L&D from the past 20 years. It is a very well organized hardware store. Large catalog of pre-written content. Users browse, select, and consume. We measure completion, enrollment, and sometimes even knowledge. We’ve been discussing delivery methods (instructor-led, self-paced, blended) to offer both live checkout and self-checkout for years. The satisfaction survey after the course was a receipt. Friendly, fast, and almost completely right next to the point of impact. And users walked out with tools and parts, but the organization had no idea what they had built with them. Or maybe they built something.
Content consumption and warehouse thinking make learning organizations look busy, but not necessarily effective. The warehouse was always measuring the wrong thing. pedestrian traffic. Catalog size. Finished product. These are inventory indicators, not construction indicators. They tell us what is left off the shelves, not whether anything of value has been built.
Rapid eLearning technology was a meaningful technology to solve specific problems of production cost and speed. But it didn’t address the deeper issue of no one being responsible for what happened after checkout.
AI is not a technology
This is where the current moment is truly different, and why it represents a paradigm shift rather than a new technological productivity boost.
Don’t treat AI as a new technology to bring to your warehouse.
AI doesn’t just help you build courses faster. This forces you to ask questions you never had to answer with the warehouse model. If we could surface the right knowledge at the exact moment someone needs it, simulate realistic conversations before they even happen, and provide personalized coaching based on real performance data, should we build courses the same way? Should we stock the shelves? How would we redefine our work? What is your output? What are your results? What are your values?
This course was always a workaround. A workaround for the impossibility of having a specialist available to every employee whenever needed. AI questions that impossibility. This means that AI-enhanced learning workarounds may no longer be needed in the same way for every problem. And L&D professionals need to consider what they were actually trying to accomplish all along. We are converting from a warehouse to a general contractor.
How is the general contractor’s approach different? This isn’t someone who stocks tools; it’s someone who shows up, assesses what actually needs to be built, sources the right resources at the right time, and remains accountable for making sure the structure functions properly upon completion. It’s a fundamentally different business model. The stakeholder relationships are different, the success metrics are different, the skills required are different, and the conversations are different.
Contractors also have problems
Here are the parts of the AI story that L&D professionals need to understand clearly: This is because getting it wrong will have serious consequences. Human contractors powered by AI can build and destroy with the same confidence, even at scale. If you let AI define you, it will.
AI contractors are very capable. They operate at speeds unmatched by humans, never tire, and can hold vast amounts of information at once. However, it has two significant weaknesses that mirror each other almost perfectly.
First, it means you’re making more mistakes than you should.
AI systems generate plausible answers with consistent authority, regardless of whether those answers are accurate or not. In L&D terms, this means that AI-generated content can contain factual errors, outdated information, or subtly misconfigured content and is delivered with the same smooth reliability as completely accurate content. You own the results. If the AI fails, it’s your fault. period. Second, it does exactly what the user asks and is optimized to satisfy the user.
AI systems are trained in a way that makes them responsive and comfortable. If you request a course, it will simply be offered to you and you will not be able to challenge the assumption that the course is the correct solution. When you ask leading questions, you get solid answers. If your brief is ambiguous, the AI will fill in the gaps with what it thinks is most likely to satisfy you, rather than flagging your brief as insufficient. Set up rules for colleagues/contributors. You need to define how the AI will work with you.
Combining these two trends could recreate the same problems that warehouses have always had (at the speed and scale of AI). Customers who don’t know what they actually need, when combined with a system designed to fulfill orders rather than question them, quickly produce a flood of confident, well-packaged wrong answers. Drywall for everyone!
What this means for L&D professionals in an AI-augmented learning environment
The general contractor metaphor clarifies what the new L&D role for AI-enhanced learning actually entails. A good contractor does more than just do what the client asks. If the brief is wrong, they will push back. They say, “You asked for this wall here, but load-bearing structures don’t work that way.” They bring professional judgment that clients don’t have and aren’t being paid to suppress.
This is exactly the skill set that L&D professionals need to develop around AI. Prompting well is really important, but beyond just prompting, it’s also important to know when to question the output, when to override recommendations, and when the AI’s confident answers are built on flawed assumptions embedded in the question.
In other words, you need to define who you are and how you want to work with AI. Don’t let the AI do it. Sure, you can sit in the back seat and enjoy the ride, but that won’t get you anywhere you want to go.
How to promote coworking?
L&D professionals who thrive in AI-enhanced learning situations are those who hone three competencies in particular.
The first is the rigor of the diagnosis.
The ability to identify exactly what performance problems actually exist before arriving at a solution, and resist both the organization’s pressure to create content and the AI’s drive to create content on demand. The second is critical evaluation.
Treat AI-generated content as a competent first draft by a junior colleague, rather than a finished product by an expert, and review it with the same scrutiny you apply to anything that has real-world impact. The third is responsibility for results.
It’s about owning the question of not just whether the content was delivered and consumed, but whether the intervention actually changed behavior.
Warehouse L&D experts were responsible for the shelving. The general contractor’s L&D experts are responsible for the building.
clearly stated opportunity
None of this takes away from what AI is capable of. For the first time, L&D has a tool that can bridge the gap between performance, behavior change, and organizational capabilities that the profession has always claimed to focus on. All of this at scale. But that gap will only close if the humans using AI have a clear understanding of its limitations as well as its capabilities. A strong contractor, where no one checks the blueprints, will not improve compared to a slow contractor. It’s a faster way to build something wrong. The current job of this profession is not to master AI. This means acquiring judgment skills that cannot be replaced by AI. Take control! Define who you are! Don’t let the AI do it.
