Guess before Google.

I’ve been saying that for years.

Before I search for the answer, I try to remember it. Before I ask AI, I try to work through the problem. Not forever. Sometimes 30 or 60 seconds is enough.

I wrote about this last year in AI as My Partner: Expanding My World Without Losing My Edge. At the time, I was mostly concerned with protecting something I’ve worked hard to develop: my ability to recall, reason, connect ideas, and make sense of things for myself.

I’m even more convinced of it now.

But I think the idea is bigger than AI.

Maybe the Friction Is the Point

There is a concept in learning science called desirable difficulty.

Robert Bjork used the term to describe learning conditions that can make practice feel harder and even make us perform worse during practice, while improving retention or transfer later.

Spacing practice, retrieving something from memory instead of immediately rereading it, and mixing different kinds of problems are examples.

That distinction matters.

Performance right now is not necessarily the same thing as learning.

Sometimes making something easier helps us perform better without helping us become more capable.

That connects with Mihaly Csikszentmihalyi’s work on flow from another direction. Flow tends to occur when challenge and skill are appropriately matched, particularly when both are relatively high. A meta-analysis of 28 studies found that challenge-skill balance had a moderate relationship with flow, although it certainly isn’t the only ingredient.

Too little challenge and we can disengage.

Way too much challenge and we can become overwhelmed.

Somewhere around that boundary between what I can comfortably do and what I can almost do is an incredibly interesting place.

That’s often where growth lives.

Have Your AI Call My AI

I joked with someone recently:

Have your AI call my AI and we’ll do lunch.

It used to be, “Have your people call my people.”

Funny, but maybe a little uncomfortable too.

Someone writes something with AI. It becomes polished, organized, grammatically correct, and probably more readable.

Then someone like me receives it and gives it to AI to help me work through it.

And I’m not convinced either person is doing anything wrong.

I’ve talked openly about having a learning disability. I can read the words just fine. What becomes difficult for me is processing and retaining large volumes of dense text. That’s an interesting combination for someone who works and thinks at a graduate level.

AI can be extraordinarily helpful there.

It can help me break down a complex research paper, interrogate an argument, translate academic language into something I can process, and then let me go back into the original source with better understanding.

That’s not necessarily outsourcing my thinking.

It can actually make deeper thinking accessible.

The difference, I think, is intentionality.

Am I using the tool to get around the difficult work?

Or am I using it to get into the difficult work?

Those are very different things.

Don’t Remove Every Hard Thing

There’s another interesting piece of neuroscience here.

I first learned about the anterior midcingulate cortex, or aMCC, listening to Andrew Huberman’s conversation with David Goggins.

If you’ve heard the interview, you probably remember it.

Huberman’s pragmatic takeaway was essentially that deliberately doing things we don’t want to do matters. Goggins, unsurprisingly, had a lifetime of experience that seemed to rhyme pretty well with that idea.

The actual science is a little more nuanced, and there is some disagreement about how far we should take that interpretation.

The aMCC appears to play an important role in how the brain evaluates effort, reward, and whether to persist toward a goal. Researchers have proposed that this region is part of the neural machinery underlying tenacity.

You’ll sometimes hear that translated into something much simpler:

Do things you don’t want to do and you’ll grow your willpower.

Maybe.

The evidence isn’t strong enough for me to state it that cleanly.

But I do appreciate Huberman’s pragmatic perspective: there may be value in deliberately maintaining some friction in our lives.

That fits remarkably well with what we know from desirable difficulty, flow, training, and frankly, lived experience.

Effort matters.

We probably shouldn’t engineer every bit of difficulty out of our lives.

That doesn’t mean making things too hard. Desirable difficulty stops being desirable when the challenge prevents meaningful learning. Even the learning literature shows that difficulties can be combined or increased to the point that they hurt rather than help.

The goal isn’t suffering.

The goal is capacity.

Assistance or Capacity?

That’s where this connects back to Comprehensive Fitness for me.

A coach can write my workout.

AI can summarize the research.

GPS can navigate.

An app can track my food.

A financial planner can build the plan.

A leader can tell me exactly what to do.

All of those things can help.

Comprehensive Fitness isn’t about proving we can do everything without help. It’s about developing the capacity to function when it matters.

Sometimes assistance builds that capacity.

Sometimes assistance replaces the rep that would have built it.

Learning to tell the difference may become one of the more important skills of the next few years.

Put It Into Practice: Guess Before Google

Try Guess Before Google once today.

When you don’t know something, don’t immediately reach for the answer.

Give yourself 60 seconds.

Retrieve it. Reason through it. Make a prediction. Write down what you think.

Then use Google, AI, a book, a coach, or another person.

Compare what you thought with what you learn.

You still get the benefit of the tool.

You just don’t skip the rep.

And maybe that’s the larger idea.

We don’t need to avoid assistance to become capable.

We need to use assistance in ways that continue making us capable.

Do Better.