Which Legos Should Educators Give Away?

I was listening to a recent episode of Lenny’s Podcast with Molly Graham when I started thinking about educators.

Molly is the person behind the “Give Away Your Legos” framework, an idea she originally developed from her experience working inside rapidly growing companies. The basic premise is that as an organization changes, you have to become comfortable handing off pieces of your job so you can grow into whatever comes next.

That is hard enough when you are handing those Legos to another person.

AI makes it weirder.

In her recent conversation with Lenny Rachitsky, Molly talks about how different it feels to give work away to a machine. These may be things you spent years getting good at and that became part of your professional identity. Now you are being asked to teach an AI how to help do them, while hearing constant speculation about whether AI will eventually replace your job altogether. (mollyg.substack.com)

As I listened, I kept thinking about educators sitting through AI training.

I have spent a fair amount of time around these sessions throughout my career. They range from basic introductions to much more hands-on sessions, so the audiences vary quite a bit.

Still, you start to recognize some patterns.

There are usually a few people who are excited. They have been experimenting with AI already, understand what it can do, and could probably lead parts of the training themselves.

Others have tried it and can see that it is interesting, but still have not found much reason to bring it into their own work.

And then there are the people who are there because someone told them they needed to be. That is often where the questions about cheating come from, along with the occasional nervous joke about AI taking their jobs.

What I rarely hear, even from the people who are interested, is a different kind of question:

Which parts of my job am I actually willing to give away?

Giving away proof that you are good at your job

The fear that AI will replace educators is easy to dismiss as exaggerated.

Most educators I meet are not genuinely expecting to walk into school next year and find a chatbot sitting at their desk.

But every so often, particularly with an experienced and technically savvy educator, I hear a more grounded version of the concern.

If this thing can do something I do faster and cheaper than I can, what stops an administrator from deciding they no longer need me to do it?

There is a lot wrapped up in that question, but the technology piece is worth sitting with.

An educator may have spent years becoming really good at a part of their craft. Maybe it is the way they build a lesson. Maybe it is the feedback they give students. Whatever it is, that skill becomes part of the evidence that they are good at their job.

So when we show them AI doing work that once required that expertise, we may be thinking about efficiency.

They may be thinking about professional value.

If the machine can do this too, what exactly am I here for?

This tension is beginning to appear in the research around AI and teacher identity as well. Recent studies have found that exposure to increasingly capable AI can increase a sense of professional agency while also increasing concerns about replacement. Other work describes educators renegotiating where their authority and professional value sit as AI becomes capable of doing work historically associated with teaching expertise. (frontiersin.org)

That sounds a lot like giving away your Legos.

I don’t actually want AI to save me time

I have a bit of a mantra about AI.

I don’t really care about saving time. I care about spending my time doing the things I want to do.

Those sound similar, but I think there is an important difference.

“Save teachers five hours a week” has become one of the standard promises of AI in education. But saving five hours does not mean much if those five hours are immediately filled with something else.

For educators, I think the better question is what they would happily stop doing.

Every educator has parts of the job that eat up attention without feeling like the reason they became an educator. Those are probably good places to start looking for Legos to give away.

What matters is what that makes room for.

Maybe an educator gets to spend a little more time with a student who is struggling. They could finally have the headspace to rethink something that has not been working in their classroom.

Or…. they just get to teach.

That means more to me than saying AI saved them five hours.

We keep starting with what the technology can do.

I think it may make more sense to start with what educators wish they had more room to do.

Some Legos are different

Of course, this raises another question.

Which Legos should educators not give away?

That feels suspiciously close to asking what makes a great educator great.

Everyone has been a student. Most of us can probably remember a teacher who was genuinely life-changing.

I doubt what made that person memorable was their ability to produce a worksheet.

It was usually something much more personal. They understood us in a way that made a difference.

Sometimes that meant noticing that we did not understand something even when we said we did. Or knowing us well enough to realize that the usual approach was not going to work.

That context builds over time.

AI can help an educator differentiate a lesson, but it only knows what we give it. It does not automatically have the history that comes from actually knowing a student, nor does it notice all the little signals that never become data in the first place.

We are simply not there yet.

That is where the educator’s value starts to look much more like judgment.

I also like another part of Molly’s framing here. AI does not need to be treated as the genius new employee who suddenly knows everything.

Right now, it may make more sense to think of it as a junior intern.

Give it something useful to do. Let it take the first pass. Then look at what it gives you and decide whether it is actually any good.

The important part is that you still need to know what good looks like.

When the work involves a learner, you also need to know whether it makes sense for that particular person.

That relationship feels very different from trying to compete with AI.

What if your job changed every six years?

There was another idea in the podcast that stuck with me.

What if your job never disappeared, but it changed every six years?

I think that is an especially interesting question for education.

Educators can have very long careers. Someone can enter the profession in their twenties and still be doing this work decades later.

Education has obviously changed throughout that time, and educators are used to change. AI feels a little different because it can change the work itself while leaving the title alone.

Tell an educator that their job could look very different six years from now and I suspect the immediate question would be, “How?”

They want to know what they should prepare for.

I don’t think any of us can answer that honestly.

Preparing for one specific version of the future is maybe the wrong goal anyway.

It might be more useful to become comfortable noticing when the job is changing and asking what that change allows you to stop doing, keep doing, or get better at.

For someone who may spend decades in education, being able to make that adjustment more than once could become an important part of the career itself.

We may be creating some of the AI burnout ourselves

This is also where I think we need to look critically at how AI has been introduced into education.

For the last few years, the message has often felt like some version of “we need to use AI or we are going to fall behind.”

For educators, that has often translated into more work.

There is another technology to understand. Student use creates new problems to deal with. The rules keep changing, and the tools keep changing even faster.

AI was supposed to reduce workload.

For many educators, at least so far, it has added another layer to it.

There is emerging evidence that AI-related technostress can affect educators’ sense of professional identity, and that stronger organizational support can soften that effect. (frontiersin.org)

That makes sense to me.

If I give away one Lego and immediately get handed three more, I have not really given anything away.

This is where organizations doing AI professional development, and the EdTech sector more broadly, have an opportunity to rethink the conversation.

Maybe we shouldn’t start with the tools

There is still a lot of “ooh, shiny” happening around AI.

Administrators can see a new capability and immediately start imagining what it might improve. EdTech companies understandably see problems they can solve.

The educator may simply see another system they are expected to learn.

That gap matters because the people buying the technology and the people living with it are not always solving for the same thing.

An administrator might hear that a product saves teachers time and see a clear benefit. The teacher might reasonably wonder where that time is actually going to go.

That is why I think AI training should start earlier than the tools.

Start with the Legos.

Before showing someone another platform, ask them what part of their work they would genuinely like to stop carrying. Then ask what they would do if they could actually put it down.

That conversation naturally gets you to the work where their judgment matters most.

Only then does it make sense to introduce AI and ask whether it can help with the handoff.

That feels very different from opening a training with five prompts that will supposedly save someone time.

There is no reason to pretend this change will always be comfortable. Molly’s original point was partly about accepting that growth requires giving things away, and her newer reflection makes the emotional cost of that even clearer. Change can create opportunity and still feel uncomfortable. (mollyg.substack.com)

For educators, I keep coming back to this question:

Which Legos am I ready to give away, and what do I want to make room for?

And for those of us building EdTech, leading schools, or training educators to use AI, there is another one:

Are we actually helping educators put something down, or are we just handing them more Legos?

A note on how this came together: I’m not much of a writer, but I do have ideas I think are worth sharing. This article was developed with the help of ChatGPT 5.6 as a thinking, research, and writing partner. The line of reasoning started while listening to Molly Graham on Lenny’s Podcast, revisiting her “Give Away Your Legos” framework, and then thinking about the educators I’ve seen sitting through AI training over the years. From there, it turned into a much bigger question about professional identity, burnout, and what it actually means to hand parts of your work over to AI. I think that process is worth being transparent about. If the idea resonates, the source of the prose shouldn’t matter. And if it doesn’t, that’s on the idea, not the tool.

The threads that informed this thinking: Molly Graham’s original “Give Away Your Legos” framework and her more recent thinking about what it means to give work away to AI; her conversation with Lenny Rachitsky about grief, burnout, and changing professional identity; emerging research on AI, teacher identity, professional agency, and technostress; and, probably most importantly, years of watching educators react to new technologies and try to figure out what those changes mean for their work. Curious what resonates, especially for educators who have already started deciding which Legos they are ready to give away.