What Everyone Gets Wrong About Reskilling for AI

In early 2023, Anthropic posted a job listing that made the news. Prompt Engineer and Librarian, $175,000 to $335,000 a year. No computer science degree required. Basic programming skills were desirable. The posting asked for a creative hacker spirit, and noted that since the field was arguably less than two years old, the role was hard to hire for.

Time magazine covered it. Bootcamps sprang up. Universities stood up certificate programs. People left steady jobs to retrain for it.

Every one of those calls was sound. The demand was real, the money was real, and for once, a leader could answer the hardest question on the table. Reskill to what? There was finally something to point at.

In 2026, Anthropic posted a similar listing. Prompt Engineer, Agent Prompts and Evals. Five years of software engineering in Python. The work is designing system prompts across products, building evaluation suites, catching regressions before model releases, and mentoring other engineers.

The salary is $320,000 to $405,000. The floor of the new range is nearly the ceiling of the old one.

The prompt engineer didn't disappear. The role climbed.

What got absorbed was the part you could teach as a short course, writing a good instruction and documenting what worked. The capabilities that once had to be assembled by hand became features of the models and the systems around them, until some firms were using AI to generate and optimize the prompts for their other AI. By May 2025, Fast Company reported the standalone role had virtually disappeared, quoting an executive who said it had turned from a job into a task almost overnight.

Nobody in that story was careless. They trained for the part of the work that could be written down, and that's the part the tools take first.

Here is the part that should worry anyone holding a workforce plan. The problem was never that companies picked the wrong skill. It's that the work hadn't stabilized yet, so the only thing available to teach was the visible behavior of a new technology.

If the work keeps climbing, reskilling can't keep chasing the job description.

The largest workforce survey on earth ran into the same wall. The World Economic Forum's 2025 Future of Jobs Report asked more than a thousand employers across fifty-five economies to name the skills their people would need by 2030. Alongside AI and data literacy, the answers came back. Creative thinking. Resilience. Motivation. Self-awareness. Curiosity.

Hold that list. We'll come back to it.

Reskilling is the agreed answer. It's in the board deck, the town hall, and even in the trade press. Upskilling the existing workforce was the most common strategy in that report, with 85% of employers anticipating it over the next five years. In McKinsey's 2025 survey, only 21% of the organizations using generative AI had redesigned any of their workflows around it. That gap gets read as a failure of will. Budgets, competing priorities, somebody dropped the ball. It's something else. You can't build a program toward a destination you can't name.

You've likely made the promise yourself. Nobody is being replaced. We want to free you up for higher-level work. Every leader has made that promise. Few can say what the better work is or how to build it into their strategy.

The Work That Replaced the Work

Start with what has changed. Recalling facts. Processing information. Analyzing data. Those abilities used to set a person apart. Now the tools do them faster than any person can, and having them is no longer an edge.

Here's what the moment demands instead. Deciding where to point the tools. Interpreting what the results mean. Questioning what everyone else has accepted. Asking what's worth pursuing at all.

That's not a forecast. In August 2026, four of McKinsey's newest partners described the change from the inside to Business Insider. The analysis layer, the work juniors were hired to do and partners earned their titles teaching, has gone to the tools.

What fills a partner's day instead is the human work. Orchestrating the meeting and deciding who belongs in it. Spotting which skill each person needs next. At McKinsey, the reskilling question didn't get answered. It got promoted.

Even Holger Hürtgen, the firm's first AI engineer in Germany and twenty years in, said he has never felt so far behind the curve, and that he's "uncomfortable every week." Teaching everyone the tool is what you do when you can't say what the new work is.

Asked what stays constant at the top of the firm, one of the four called it an irreducible core. Another named it. Curiosity.

Now go back to that list from the Forum. Same word. Among the skills rising fastest, the report separates out two it calls socio-emotional attitudes, and curiosity is one of them. A thousand employers were asked to name the skills their people would need by 2030, and the answer they keep reaching for is something the report itself won't call a skill.

Curiosity is the most honest answer the language currently offers, which is why the smartest people in the room keep landing there. It's real, and it's not enough. Curiosity is the impulse to ask. It doesn't build the answer.

Imagination builds the answer.

The tools can generate a hundred options and never know which one matters. Choosing is the human edge — reading what the data means, seeing what's missing, deciding what's worth building. Almost no one calls it by name, because it has always been treated as a gift you either have or you don't. That's what my research changes. Imagination has a structure, and structure can be learned.

Where should your organization focus first?

Follow the claim and it changes what performance means. If imagination is the competence this era demands, it belongs where competence gets defined, in the review, in the promotion, in what doing the job well means. People develop what they are measured against. Leave imagination off the criteria and no amount of training will produce it.

Nothing Is Missing

No organization is short on imagination. It built everything you have. The question is which direction it runs.

My research into how the mind builds futures found three modes, and my Imagination Framework is how they work together. Navigational asks how do I get there. Protective asks what could go wrong. Generative asks what could be, the only one that pictures something that doesn't exist yet.

Here is why the Framework matters now. AI accelerates the first two modes. Planning gets faster, defense more efficient, execution more precise. Generative sits outside what any tool does. A model recombines what it's given and never wonders what's missing. It's also the mode that shuts down first under pressure. The tools amplify the two your organization already overuses and leave untouched the one it has stopped practicing.

Why did Generative shut down? Not for lack of capacity. The other two modes are busy protecting the Default Script, the definition of success that was correct once, has been confirmed by every quarter since, and now goes unquestioned because questioning it feels like questioning the company. It's why the workforce plan came back as a curriculum. Navigational picks the training, Protective covers the risk of falling behind, and nobody in the process ever asks what the work is becoming.

Every Function, Same Question

This is not a call center story. The freed up time is real and it's everywhere. In BCG's 2026 survey, 42% of frontline AI users said it saves them a full day a week. Two-thirds get little or no guidance on what to do with it. The time is there. The decision about what to do with it is not.

The work built most tightly on defined procedure went to the tools first. Every function has some of that work. Engineering, legal, operations, sales. The routine work shifts, capacity opens up, and somebody has to decide what it's for.

This is the hole in most AI strategies. The tools get implemented, the time comes back, and no one decides what to do with it, so the Default Script decides instead, and the old plan absorbs everything the tools give back.

Now What

Here's what everyone gets wrong about reskilling for AI. It's treated as a training problem. Pick the skills, buy the curriculum, run the program. But a program can only teach what's already specific enough to teach, and that's the first thing the tools take.

Stop trying to predict the next job. Build the capability that lets people move when the job moves.

Anthropic's prompt engineers moved up a layer in three years. The more precisely you can name a skill, the shorter its clock.

Your question is no longer which course to buy. It's where your organization's imagination went, and which definition of success it's defending while the freed up time disappears back into the old plan.

Where should your organization focus first?

Drawn from my book, The Imagination Age.

What Your Best Thinking Is Defending

On the morning of September 11, 2001, call volume in New York surged to thirteen times normal and the cell networks collapsed. Millions of people couldn't reach their families. The exception was anyone holding a BlackBerry because it ran on its own data network, separate from the cellular grid. Government agencies took note. Within a few years the federal government relied on the device so heavily that when a lawsuit threatened to shut it down, the Justice and Defense Departments went to court to keep it running. When President Obama reached the White House, he wouldn't surrender his BlackBerry, even over the objections of his own security agencies. It was that beloved. A device for bankers had become something else. It became infrastructure.

Revenue climbed more than sixty percent in a single year. Subscribers went from one million to fourteen million in four years.

That kind of growth rested on a single belief. A phone was a serious instrument for serious people — built for security, reliability, and a keyboard fast enough for real work. Wall Street believed it. Washington believed it, and seventy-seven percent of federal managers carried one. The market believed it too. By 2009, BlackBerry held more than half of all smartphones sold in America, and Fortune named it the fastest-growing company in the world. A belief this widely shared stops looking like a belief at all. It looks like reality.

I loved mine. I have fond memories of dashing between meetings in New York, tapping out emails with one hand. I believed it too.

You already know how the story ends. That's the part hindsight hides. BlackBerry was right for a very long time.

On January 9, 2007, BlackBerry's co-CEO, Mike Lazaridis, was on his treadmill when the report came through that Steve Jobs had unveiled the iPhone. A former engineer, he understood what he was seeing. Apple had put a full web browser on a phone, a data load the networks had always refused to carry. Somehow Apple had gotten one to say yes. In a business where the carrier controlled the device, the software, and the customer, Apple had demanded the reverse, plus a cut of every monthly bill. Verizon walked away. AT&T took the deal. The next day Lazaridis pulled his co-CEO in front of a screen to watch the announcement. "These guys are really, really good," he said. "This is different." McNish and Silcoff report the scene in Losing the Signal, their account of the company's rise and fall.

His co-CEO, Jim Balsillie, felt confident, as he had every reason to be. "It's OK, we'll be fine."

Larry Conlee, the company's chief operating officer, studied the device and saw exactly what was wrong. The iPhone wasn't secure, not by the standard that had made BlackBerry the government's phone. Its battery drained fast, and its glass keyboard was miserable. The mobile web was a market BlackBerry never wanted. By every measure the company used to judge a phone, the iPhone came up short. They weren't wrong. These were the most knowledgeable people in the industry, and they read the flaws with clear eyes. What they missed was bigger. Apple had not built a better phone. Apple was not a phone company. It came from computers and music, five years into the iPod, and carried its own idea of a handheld. Apple was asking what a phone could be. BlackBerry had stopped asking.

It wasn't a lack of intelligence. They were certain, and that certainty was the work of their own imagination.

The same imagination that had pictured wireless before the world caught up was now locked into what BlackBerry already knew. It ran through every reason the iPhone would fail, brilliant work pointed the wrong way. No one asked what a phone was becoming. They were too busy defending a success the whole market had proven right.

Same engine, different direction. Generative imagination, the mode that pictures what doesn't exist yet, had shut down, and Protective imagination, the mode that scans for threats and guards what already works, took over. Under pressure, a brain defaults to defense.

This Is the Default Script

The Default Script is the definition of success an organization optimizes toward and never examines. The definition doesn't have to be succeeding. It only has to go unquestioned. When the script is proven, it sinks so deep into how the organization allocates capital, sets targets, and promotes that it stops reading as a strategy and starts reading as the way things are. It doesn't take a collapse like BlackBerry's. Most scripts are smaller and harder to see, running at the level of a team, a leadership group, the customer experience, or the whole organization. It isn't a bad plan. BlackBerry's was a spectacular plan. That's what makes it dangerous.

A bad plan gets argued down. A proven one recruits the company's best thinking to its own defense, and that defense is strongest when the script is most correct. The pattern is measurable. When researchers tracked the airline and trucking industries through deregulation, they found that the more success a firm had behind it, the longer it held to its old strategy, and the more ground it lost. BlackBerry's was never more correct than in 2007. That was the most dangerous moment it would ever have, and it felt like the safest.

This is not the exception. Half of the S&P 500, the largest public companies in the country, will be replaced within a decade. The ones that fall are rarely the companies that were failing. They are the ones that were sure, running on what Innosight calls a confidence bubble, a readiness they assumed and never built.

You Don't Run BlackBerry

You are not running a company in freefall. Your numbers are strong, your team is deep, your board is largely aligned. That is where the script thrives. It does its most durable work inside organizations that are performing well, and something in the BlackBerry story is already familiar.

The oldest institutions carry the deepest scripts. A century-old company, a John Deere or a major bank, is not headed for BlackBerry's ending. Its script moves slower, setting the terms of what gets proposed, what gets funded, and what the market is allowed to look like, inside the company and beyond it. Questioning a script that has been right for a decade is hard. Questioning one that has been right for a century is close to heresy.

The strategic review keeps circling the same short list of options, every one reasonable, and nothing ever changes. The big innovation bet takes eighteen months and real money and returns almost nothing. The offsite produces a slate of initiatives any competitor could write. The board approves the AI investment in a single meeting and bolts it onto the existing business model, so the company only gets faster at the strategy it already has, right or wrong.

None of it looks like decline. Every meeting is sharp, the people are the best you have, and that is the tell. The work never gets worse. It stops reaching for what could be next, and from the inside that counts as a good year. In a bad year, it hides just as well, and the losses get blamed on execution, so the answer is always to run the old plan harder. Good year or bad, no one questions the premise. By the time you feel the cost, the market has shifted and the script has not.

The reason it can't be read from inside is structural, and it falls hardest on the person most equipped to lead. The executive who rose by mastering the model is the last vantage from which it can be seen. The organizational researcher Erik Dane calls this cognitive entrenchment. As expertise deepens, the patterns that produced it harden, and the expert loses the flexibility to see past them. To them the script doesn't feel like a script. It feels like knowing what works. The cost is personal before it is institutional. The leader most fluent in what already works spends their strongest years getting better at it as the ground shifts underneath them.

You can watch it happen when someone new walks in. They ask the questions an outsider still can, why the pricing is built the way it is, why that market never comes up, and the room answers, swiftly, that it knows what works by now. Listen to that answer. It’s more dangerous than “we've always done it this way”, because the older phrase at least admits it is describing a habit. This one believes it is describing the truth.

The script is not new. What is new is the speed. When MIT studied enterprise AI in 2025, it found ninety-five percent of corporate pilots returned no measurable impact to the bottom line. AI does not choose the target. The script does. Point it at a definition of success no one has examined, and the company reaches the wrong place faster.

The Script Is a Decision

Escaping it takes imagination, the one faculty the script claimed first. Even clear sight is not enough. Lazaridis saw the iPhone for what it was and said as much, and nothing changed. A script doesn't live in any one head. It lives in what the organization rewards, what it measures, what it waves off, and a mind rehearses what earns approval until it feels true.

It runs all the way up. A CEO reports against the plan the board approved, and the board grades the CEO against that same plan, so neither one has a reason to question a number that keeps coming in on target. The machinery built to protect the company ends up guarding the script.

The script keeps its power by passing for reality, and no one argues with reality. But look closer and it is something smaller. Someone made a decision, years ago, for reasons that held up at the time. A decision can be revisited. The organization doesn't have to move a single thing first. The seeing shifts, and then the script can.

Go back to the two moments. September 2001, when BlackBerry became the phone that held while everything else failed. January 2007, when the iPhone arrived and BlackBerry saw no threat. They were the same event. BlackBerry had defined the market it led, secure, reliable, built for people with serious work to do. That definition became the script, and every year the numbers came back and confirmed it. That was the trap. The engineers studied the iPhone closely and read it correctly against a script they had written themselves. Being right the first time is what they never thought to question. Balsillie said they would be fine, and by the terms of the script he was right. The script had been sound once. As the ground shifted, no one asked whether it still fit.

That's the question underneath the BlackBerry story, and it isn't how to avoid their fate. It's simpler, and it comes first. What is the thing your organization is most certain about. That's the thing you've stopped questioning. It was probably right once. So was BlackBerry's.

*

The Default Script DiagnosticFive minutes. See where your organization's script has gone unquestioned.

Work with TheresaFor leadership teams ready to uncover their script and start to shift it.

The Imagination AgeThe three modes of imagination and the disciplines for turning them back toward the right question.

Very Good at the Wrong Question

Best Buy's new CEO skipped headquarters on his first day. He drove sixty miles north of Minneapolis to a store in St. Cloud, Minnesota, put on a blue shirt, and worked the floor.

It was September 2012, and the company was widely expected to die. Circuit City had run the same playbook and lost. It liquidated in 2009, all 567 stores dark inside a few months, the second-largest electronics chain in the country gone. Borders followed in 2011, the last of its bookstores shuttered by that September. Two big-box chains gone in three years. Best Buy looked like the next one to fall.

The stock was trading near a ten-year low. The founder was assembling a bid to take it private. Wall Street had written the obituary, and the analysts agreed on the cause of death. The stores. Too many, too expensive, too easy to walk into and then buy the same television from Amazon on your phone before you reached the parking lot. Stores were the problem. The only argument was how many to close, and how fast.

Hubert Joly worked the floor for three days. Over lunch he asked the store's general manager to sketch the layout on a napkin. A fifth of it went to physical media, already dying. Mobile phones, the one category booming, held just four percent of the space. At the back of a sixteen-billion-dollar small-appliance market he found a single blender. One.

The analysts saw a cost to cut. Joly, on the floor, saw a building full of decisions no one had revisited in a decade.

You may be sitting in a version of that room right now. The strategy keeps coming down to the same two options — two answers, two sets of data, and a meeting that ends where it began. You have read the deadlock as a strategy problem, and you are building the private case for whichever option you trust more. Hold that thought. It is the same room.

The analysts were not careless. They were some of the most capable people in retail, and they had run the numbers the same way for years. The stores cost money. Online was cheaper. Amazon had no floors to heat or staff to pay, and every quarter the comparison got worse. Inside that arithmetic, closing stores was not a position. It was what the data said.

This is what a settled binary looks like from the inside. Two options, weighed against each other for so long that the argument becomes the whole field of vision. The Best Buy version was existential. Most versions are smaller. Build the feature or buy the company. Premium or volume. Sunset the product or double down. Centralize or hand it to the regions. The hardest versions are about people. Cut the team or hold steady. Five days in the office or fully remote. The shape repeats at every altitude, and the danger is never that one of the options is wrong. The danger is that these may not be the right questions at all, and an organization can get very good at answering questions it stopped examining years ago.

That frame did not come from a meeting. It came from an inherited definition of how the industry wins, built in an earlier era and absorbed long before anyone in the room took their seat. Big-box retail ran what The Imagination Age calls a Default Script: scale the footprint, drive foot traffic, win on selection and price. The people who rose inside that world rose because they executed the script well. The metrics rewarded it. The promotions confirmed it. By the time the ground shifted, the script had stopped looking like a strategy and started looking like the shape of reality.

A script that strong does more than guide decisions. It decides what the team is able to perceive. The options that fit the script are visible and arguable. The option that sits outside it does not register as a worse idea. It does not register at all. If you have walked out of a strategy session with every option on the board reasonable and not one of them new, you have felt this from the inside.

Why the Room Can't See It

Why couldn't a room of experts see in two years what Joly saw on the floor in three days?

The answer is more physical than it sounds. In a 2005 study in Biological Psychiatry, the neuroscientist Gregory Berns gave volunteers a perception task. They looked at pairs of three-dimensional shapes and judged whether the two were the same object rotated or genuinely different. The task had a right answer. Before each person decided, they saw how others had answered, and sometimes those others were wrong. People often abandoned the correct answer and went along with the group. The revealing part was where the going-along showed up in the brain. When the wrong answer came from other people, activity changed in the network that does the seeing, the visual and spatial regions running the task, and not in the regions that handle deliberate choice. When the same wrong answer came from a computer, that shift was weaker. The researchers drew the unsettling conclusion. Social pressure was not overriding people's judgment after they took in the shapes. It appeared to reach back into perception itself, shaping what they saw before judgment began. Conformity, on this reading, was not caving on a belief. It was closer to seeing what the group saw.

Four years later, Vasily Klucharev, writing in Neuron, found the enforcement mechanism. When a person disagreed with the group, the brain produced the same error signal it fires when you simply get something wrong, and at the same moment pulled back the small reward that comes with belonging. Disagreement registered as a mistake and a loss at once. Berns had found the third piece earlier. Hold your ground against the group, and the amygdala, the brain's threat alarm, fires harder, treating independence as a kind of danger.

Put those together and a team's agreement stops looking like weakness. The brain pulls what you see toward the group. It treats disagreement as an error. It withdraws the reward of belonging the moment you break ranks, and sounds a threat alarm on anyone who stands alone. Four systems, pulling the same direction, none of it reaching awareness. Convergence is not what a team does when it loses its nerve. It is what the brain does on its own.

That was the Best Buy boardroom. A roomful of retail experts looking at the same stores, able to see only what the room had already decided was there. It was also your last leadership meeting, in every head at the table, yours included.

Where the Imagination Went

Here is what was happening to that team. Imagination runs in three modes — the framework at the center of my book — each doing a different job. Generative asks what could be. Navigational asks how we get there. Protective asks what could go wrong. A binary only ever exercises two of them. Choosing between options that already exist is Navigational and Protective work, planning the route and bracing for the risk. The mode that pictures a third option, the one not on the table, is Generative, and it is the first to shut down when the pressure rises. The endless either/or in your meeting is not a sign that your people have stopped imagining. It is the signature of a team running on two of its three modes.

The imagination never left the room. Same engine, different direction. A transition puts the team under threat. A merger, a collapsing market, a new leader with a mandate to fix things. The brain reads all of it as danger and runs Protective. A team's collective imagination goes to work guarding the existing frame instead of breaking it. It builds the case for each option. It anticipates every risk. It defends the script better than it ever defended a new idea. Because consensus has already rewritten what everyone sees, the guarding feels like nothing more than facing facts.

He Had to Leave the Building

This is why Joly's first move was the floor.

He was not gathering folksy anecdotes. He was doing the one thing he could not do from inside the company's own consensus. He took in information the room had not already filtered. He later said the things he learned standing in that store were things he could never have seen poring over spreadsheets or sitting in meeting rooms with other executives at headquarters. The napkin, the dying media wall, the lonely blender, the phones booming in four percent of the space, none of it was visible from inside the room, because the room could only see the stores the script had defined. Joly needed new inputs to bring the Generative mode back online. Then he had to do the harder thing, and point it at what could this be while every analyst and every instinct screamed what could go wrong.

What he saw was that the stores were never the liability. They were the asset Amazon could not build. He turned them into mini-warehouses, shipping online orders from shelves that already sat near every customer. He rented floor space to Samsung and Apple and Microsoft, who paid to staff their own showrooms inside his. He matched Amazon's prices in the aisles, so the showrooming that was supposedly killing him became the reason to buy on the spot. None of this split the difference between cut and keep. It redrew the question. A company that had lost more than a billion dollars turned a profit, and kept turning one. The asset was hiding inside the thing they had agreed to kill.

Everything that blinded that team is running in yours. The pull toward what the group sees, the alarm on whoever breaks from it, the script deciding what counts as a real option before anyone speaks. Seniority does not lift you out of it. The longer you have succeeded inside the consensus, the harder it holds you, because you helped build it.

This is why the third option stays invisible. It is not missing. It is sitting in plain view of someone who is not in your meeting, in the exact way a booming phone market was invisible to a roomful of experts until one of them stood in the store. Generative imagination does not switch back on by trying harder in the same chairs. It comes back when new information reaches the room from outside the consensus that filters it.

That part is yours to start. You do not have to run the company to do it. You have to be the one who brings in the voice the room has never had to answer to, the one who treats the two options on the table as a question worth reopening. Someone has to carry in what the room cannot see on its own.

Joly put on the blue shirt.

Talk to Strangers

Colgate-Palmolive couldn't figure out how to get fluoride powder into a toothpaste tube without it scattering everywhere. Its own chemists were stuck. The company did something that should have been absurd. It posted the problem to the open internet and offered a reward to anyone who could solve it.

Not consultants. Not a rival lab. Anyone, in any field, anywhere in the world, through a platform called InnoCentive, built by the drugmaker Eli Lilly for exactly this kind of problem.

A Canadian engineer named Ed Melcarek solved it. His answer was to put a positive charge on the powder and ground the tube, so the particles flew where they were sent, basic physics that Colgate's chemists had never thought to apply. He was paid twenty-five thousand dollars. He did not work at Colgate. No literature search, no recruiter, no one inside the company would ever have found him.

InnoCentive has run this thousands of times. Researchers studied 166 scientific problems from the labs of twenty-six companies, attempted by more than twelve thousand solvers. Roughly a third of the problems that had defeated experienced corporate scientists were solved by outsiders. One especially stubborn problem drew solutions from a carbohydrate researcher in Sweden, a retired aerospace engineer, and a veterinarian, not one of them in the field. And the further a solver's expertise sat from the field of the problem, the more likely they were to solve it.

Not as likely. More likely.

The Expert's Groove

This runs against the premise every organization is built on, that the person deepest in a domain is the one most able to solve its problems. The InnoCentive data says the reverse. The chemist working the chemistry problem carries a disadvantage the aerospace engineer does not.

The disadvantage has nothing to do with how much imagination the expert has. The expert has spent decades imagining inside the field, picturing how molecules will behave, running scenarios a novice couldn't follow. The trouble is direction. That imaginative power has been pointed one way for so long that it now runs in a single groove. Show the mind a new problem and it reaches for the pattern that solved the last one, and the answer forms before the question has been examined. This is the cognitive entrenchment described in What Gets Harder When AI Gets Better, the wall that expertise itself builds. The novice has no familiar pattern. The expert reaches before seeing the problem has changed.

The outsider has no groove. The aerospace engineer looking at a chemistry problem cannot route it through twenty years in the field, because there are no twenty years. There is only the problem and a set of instruments shaped by an entirely different world. The engineer's imagination builds what the specialist's can no longer build, the thing the field had trained its experts to stop seeing.

What Imagination Is Doing

When you picture something that isn't in front of you, a solution, a strategy, a conversation that hasn't happened, your brain runs a coordinated process across three networks. The default mode network builds the scene, stitching fragments of memory and knowledge into a plausible new construction. The executive control network evaluates what the default mode produces. Is this realistic? Useful? Worth pursuing? Between them, the salience network decides what deserves attention and when to hand control from free construction to hard evaluation. When the three cooperate, you can drift into an idea, test it, and move on it with confidence. When stress or fatigue or pressure knocks them out of sync, the construction either drifts without aim or shuts down before the idea takes shape.

When neuroscientists Eleanor Maguire and Demis Hassabis mapped this system, they confirmed that the regions the brain uses to remember the past are the same ones it uses to imagine the future. Memory isn't a recording. The hippocampus binds fragments of experience so the mind can recombine them, letting you walk through a remembered kitchen and an imagined boardroom with the same fluency. Imagination is memory turned toward what hasn't happened yet.

Roger Beaty and his colleagues found that highly imaginative people show stronger connectivity among these networks. Connectivity, like any pattern in the brain, rewires with use. Neuroscientists call this neuroplasticity, and it means the capacity is not a fixed trait handed out at birth. It strengthens through practice, the way a muscle strengthens.

The expert hasn't lost the capacity. The networks are intact. They've been running the same two jobs for so many years, planning the path and defending what works, that the third job, building what doesn't exist yet, has fallen out of practice. The stranger has no special talent for this. The stranger's networks simply aren't grooved on your problem, so they build freely where yours build on rails. The capacity rebuilds with practice. The only question is where the practice happens.

And With Whom

The earlier piece, What Gets Harder When AI Gets Better, ended right here, on the claim that the competence rebuilds with practice and not in the rooms that caused the atrophy. It left the harder question open. A reader answered it with an analogy. AI is like handing everyone the keys to an airplane with no flight training and no air traffic control. The training is individual capacity. Air traffic control is the collective frame. Most of the AI conversation treats the engine as the whole problem and skips the question of where we're flying, and with whom.

And with whom. That is the part that goes missing, and the InnoCentive results show that who's in the room decides everything.

Rebuilding imagination gets framed as a personal exercise. Protect time on the calendar. Ask a different question. Catch the dismissed idea before it leaves the room. All true, all necessary, and all of it still assumes the room is yours, your team, your industry, your building, the people who share your training, your blind spots, and your standing. That room is the problem the InnoCentive results expose. Put a group of experts from one field on a problem from that field, and the room runs in a single groove, multiplied by everyone sitting in it. The same pattern gets reached for around the table. The same thing gets missed. Conviction rises while the range of what anyone can picture stays exactly where it started.

The solo answer fails for the opposite reason. Thinking alone frees you from the room's shared blind spot but starves the imagination of the raw material other minds supply. The default mode network builds scenes out of what it has. Give it nothing new and it recombines the same fragments. Neither the room of your own people nor the desk by yourself is where the competence rebuilds.

The Two Things the Room Has to Do

What rebuilds the competence is a room that does two things at once, and they pull against each other. Range wants strangers. Safety, in most rooms, comes from people who already know you. Chase one and you lose the other.

The first is range. Minds from outside your field, whose imaginations run in grooves yours never cut, able to build the scene you have lost the ability to build. InnoCentive got this through distance, broadcasting one problem to thousands of strangers who never met. A room gets it through proximity, putting the outsider across the table from you. Either way the principle holds. The point is accuracy more than variety. The outsider is the one positioned to tell you the problem on the table is the wrong one.

The second is harder to build, and most rooms get it wrong. The room has to be safe enough that people will think out loud. Comfort is not the same thing. The safety has to be structural, where no one present can grade you.

Across decades, research on group idea generation has found that people produce fewer original ideas under evaluation, and the effect compounds when the evaluator holds power over them. The expectation alone is enough, with no one saying a word of criticism. Put a senior leader in a room with peers who will remember, reports who are watching, or a board that controls their future, and the salience network reads threat and hands control to the protective mode, the one built to defend what already works. The default mode network never gets to build. The generative work shuts down before the conversation starts. The leader performs certainty. They will not risk the unfinished thought. And the unfinished thought is the raw material imagination needs.

This is why the room that rebuilds imagination cannot be your own organization. Your organization is the one place you can least afford to be seen building badly on the way to building well.

A Room Built on Purpose

Pixar's Braintrust exists to take an unfinished film apart. Smart people in a room, finding everything wrong with work that isn't working yet. It functions because of two rules its founders made explicit. The people giving feedback hold no authority over the project, and the director can use all of it or none. And they never prescribe the fix. They name the problem and leave the solution where it belongs. Strip the authority out and honesty becomes possible. The moment feedback carries power, the threat response takes over and candor gives way to performance. The most senior people in the building could hear the hardest things about their work, because no one in the room held power over it. They credit the Braintrust with a run of films that doesn't happen by accident.

Pixar solves the safety half. The range half comes from rooms built across fields, the way InnoCentive built one across thousands. Put both halves together and you have the room imagination needs. Minds from outside your field, in a space where none of them can grade you. Range and safety, at the same time, on purpose.

That room is close to impossible to assemble for yourself. The calendar won't produce it. Your network mostly returns your own field. The people who would make it work are the ones you don't yet know, in industries you don't operate in, with no stake in your standing.

The Work That's Still Yours

What gets harder when AI gets better is the generative work — picturing what isn't there yet, deciding what deserves to exist, recognizing when the question itself is wrong. That work is imagination, the human competence this era demands. It rebuilds with practice, and the practice needs a room. The room that works is the one your instincts and your calendar will never build, because every instinct says to solve the problem with the people standing closest to it.

The leaders who rebuild this competence will stop trying to think their way out alone, and stop relying on the only room they already have. They'll go find the strangers. They'll make the space where no one can grade anyone. And they'll prove for themselves what the InnoCentive results have said all along, that the person who can finally picture your problem clearly is almost never the one who has been staring at it the longest.

Colgate's chemists stood closest to the problem. A stranger solved it with physics.

---

What Gets Harder When AI Gets Better

In 1903 two teams were racing to achieve powered flight. One had a Congressional grant and the most powerful aircraft engine ever built. The other had a bicycle shop in Dayton, Ohio.

Samuel Langley, Secretary of the Smithsonian Institution, was the most credentialed aeronautical scientist of his era. A Congressional grant of $51,000 (roughly $1.9 million in today's dollars) funded his work on the Aerodrome, powered by a 125-pound, 53-horsepower gasoline engine designed with his assistant Charles Manly.

The Aerodrome was launched from a catapult atop a houseboat on the Potomac in October 1903. With news reporters watching, it promptly collapsed on itself and fell into the water. A second attempt on December 8 produced the same result. As Smithsonian Magazine recounts, the press was unsparing. The New York Times ran "Flying Machine Fiasco." The Washington Post called the craft "a total and admitted failure."

Less than ten days later, Wilbur and Orville Wright achieved sustained powered flight at Kitty Hawk. They had built the Flyer for about $1,000 (roughly $38,000 in today's dollars), funded from their bicycle shop in Dayton. They constructed a homemade wind tunnel and asked a different question. The unsolved problem was how a pilot would balance an aircraft against unstable air. Their three-axis control system is still the foundation of every aircraft built since.

Langley solved the wrong problem with extraordinary precision. The Wrights had spent years watching how birds held themselves steady in unstable air, and imagined a flying machine that could do the same.

AI is the Langley engine of this moment. It will optimize, model, and forecast at a power no leadership team has had before. It cannot tell anyone when the engine is pointed at the wrong problem.

Sixty percent of US CEOs say changing the business model is their top priority for boosting profitability this year (Conference Board). PwC's latest CEO survey adds two more numbers to the picture. Just thirty percent of CEOs are confident about revenue growth, the lowest reading in five years, and only twelve percent say AI has delivered both cost and revenue benefits. Pressure is up, conviction is down, and the space between them is what senior leaders are now being asked to close.

AI was supposed to close it. Three years after it moved to the top of every leadership agenda, the senior leaders who see what's getting harder are keeping it to themselves, and the conversation that could change the trajectory isn't happening.

What gets harder when AI gets better.

AI is consolidating the work senior leaders have excelled at for decades. What remains is the harder work of picturing what isn't there yet and deciding what deserves to exist. Most are out of practice.

The Reinvention Problem Is Not a Strategy Problem

Every offsite is a reinvention offsite now. The board has been explicit. Original strategy, original positioning, something a competitor hasn't considered. Several rounds in, with AI threaded into every transformation memo, the plan has become indistinguishable from what every competitor is producing. Sharper language, updated metrics, same trajectory.

The pattern shows up anywhere a senior leader is being asked for a move no one has made yet. It's the sixth version of the investor pitch the audience has already heard, the annual letter that could have been written by three peer foundations, the product roadmap that mirrors every competitor's, the marketing that reads like every other CMO's. Same outcome, different rooms.

CEOs know the model has to change but don't believe they're moving fast enough. PwC research finds the top third of companies that reinvented their business models outperformed industry peers by 71 percentage points on a combined measure of profit margin and revenue growth. The advantage isn't from better execution. The companies that fall behind tend to be the most disciplined of all, running playbooks with precision against a picture that no longer matches the terrain.

Two more signals never surface. The senior leaders the institution is counting on for the next move have started pulling back. They're still in their seats, but they've disengaged in a way that engagement surveys won't catch for six months. The one accountable for what comes next is sitting with an unsettled feeling. The instinct is to dismiss it. Not one of those offsites produced a move a competitor couldn't replicate.

The feeling is signal, running ahead of what the surveys and the market will eventually confirm.

AI is consolidating the cognitive work senior leaders have been rewarded for throughout their careers — pattern recognition, optimization, risk modeling, scenario planning, and forecasting. Machine learning now matches or exceeds human performance on most of it, and the consolidation is accelerating.

What's left is what AI cannot do. That work belongs to imagination as a competence. Sensing what's missing, choosing what to commit to, reading what the data can't show, recognizing whether the question on the table is even worth asking.

That cognition has been atrophying in senior leaders. The work AI now handles is the work the institution trained them to do, promoted them for doing well, and reinforced at every level of seniority.

AI is consolidating what the institution rewarded. The work that's left was never cultivated with the same intent.

The Atrophy Is the Diagnosis

This is not a willpower problem. It's how expertise works.

Erik Dane's 2010 paper in Academy of Management Review makes the mechanism specific. As people develop deep expertise in a domain, their thinking gets locked inside the schemas of that domain. Dane calls it cognitive entrenchment. Domain mastery produces inflexibility in problem-solving, adaptation, and creative idea generation. The deep specialization that makes a leader excellent at one way of thinking entrenches the schemas for that approach, while the ones not in use grow less accessible. The competence that built the authority becomes the wall against everything else.

What makes the entrenched expert's blind spot distinct is the confidence behind it. The novice gets stuck because nothing looks familiar. The entrenched expert gets stuck because everything does. A novel situation activates the schema that solved the last one, and the expert responds before examining the question.

This holds across fields. It holds for the partner who has won the same kind of case for fifteen years. It holds for the operator who has scaled the same playbook through three companies. It holds for the founder whose pattern recognition built the company and now blocks them from seeing what comes next.

Imagination is the competence this era demands. It operates in three modes. Navigational plans the path forward, optimizing, sequencing, and executing. Protective scans for threat and defends what's working. Generative pictures what doesn't exist yet and decides what deserves to exist. Same engine, different directions.

The first two modes work inside an accepted frame. Navigational asks how to get there. Protective asks what could go wrong on the way. Both presume the destination is correct. Generative interrogates the destination itself. It's the mode that tests whether the right question is on the table.

AI accelerates Navigational and supports Protective. Generative is beyond its reach.

Generative fades first under sustained pressure. The institutions that shaped most senior leaders rewarded the other two for years, anchored to quarterly cadences and risk dashboards, promoting careers built on optimization and never being wrong. The leader can still execute and still defend, but Generative has been crowded out by everything the other two demanded.

The muscle hasn't disappeared. Atrophy describes what happens when a competence stops being used. It rebuilds with practice, and the practice doesn't happen in the same rooms that caused the atrophy.

Recognition Before Practice

What rebuilds the muscle is small, deliberate practice on unfamiliar ground.

Recognition comes first. The leader who can name which mode is operating can change it. The one who only feels stuck can't. Most leadership failures right now come from running the wrong mode. The moment needed Generative, the room defaulted to Navigational, and no one had the framework to name what was happening.

The work after that is unglamorous. It looks like protected time on the calendar that isn't optimizing, defending, or executing. It looks like a different set of questions in the strategy session, the discipline of catching the dismissed idea before it leaves the room, and practice with other senior leaders sharpening the same question against different contexts. The muscle rebuilds faster when the thinking happens outside your own industry, your own habits, and your own defenses.

This is the work senior leadership now demands. None of it will show up in the metrics that get tracked.

Vision Isn't a Trait

The directive coming down is the same everywhere. Be more visionary. Bring more vision to the strategy. Help us see around the corners.

The ask is clear, but the muscle isn't there. Senior leaders take the note, sit down to do something with it, and find that nothing moves. Trying harder won't help. Vision is what imagination yields in Generative mode, the output of a competence that hasn't had reps.

What gets harder when AI gets better is the work that was always human, and was always the first to fade under pressure. Senior leaders who recognize this and rebuild the muscle will define what comes next. The ones who don't will keep optimizing what they're already doing — faster, against the same picture as everyone else.

The work that was always human is what imagination does. That's the work that's still yours.

The three modes, the Generative Cliff, and the disciplines for rebuilding the muscle are all in The Imagination Age.

---

The Life You Didn't Choose

You're good at your life. That's not the problem.

You hit the milestones ahead of schedule. You can articulate why you made every major decision, and the reasoning holds up. The degree led to the career. The career led to the house, the partner, the account balance that lets you sleep at night. People tell you you're doing well. You believe them. Most of the time.

But there's a thing that happens on a Sunday night, or in the car after a dinner party, or in the ten seconds after you close your laptop and before you pick up your phone. A gap. Not unhappiness. Something stranger. The sense that the life you built is working perfectly and that you can't remember when you last felt like it was yours.

You file it away. You're busy. The week resets.

That gap has a source.

What It Is

The Default Script is the set of assumptions about what a good life looks like that you absorbed before you had the language to question them. What counts as achievement. How fast it should happen. What a person your age, your education, your tax bracket is supposed to want next.

It wasn't written by anyone with bad intentions. It was written by people who loved you, cultures that shaped you, and systems that rewarded you for staying on the path. Get a degree or two. Find stable work. Build wealth. Settle somewhere respectable. Retire once you've earned the right to rest.

Most of us don't argue with the script. We don't even recognize it as a script. We think it's just how life works.

The claim here is specific, and it's more uncomfortable than it sounds. The Default Script didn't just hand you a path. It handed you your desires. You didn't just follow someone else's plan. You were trained to want it. The wanting itself was installed.

How It Happens

Three mechanisms. Each one is well-documented. And each one closes a door you probably assumed was still open.

You watched, and your brain took notes.

Psychologist Albert Bandura called it social learning. Before you ever made a decision about what kind of life to build, you'd spent years observing which choices got rewarded and which ones drew worry. You watched which adults seemed calm and which seemed afraid. You watched what happened to the cousin who went to law school and what happened to the one who moved to Austin to paint. Nobody sat you down and explained the rules. The rules were in the atmosphere. Your brain cataloged them before you knew it was keeping score.

By the time you're choosing a major or accepting a job, the architecture is already built. The options that feel "realistic" are the ones that match patterns you absorbed years ago. The ones that don't match feel risky, impractical, self-indulgent. Not because they are. Because the prediction machine classified them before you got to weigh in.

This closes the first escape hatch. "That's just how I was raised" isn't a neutral fact. It's a description of the installation process.

The first number became the only number.

Amos Tversky and Daniel Kahneman identified something they called anchoring bias. The first standard you encounter becomes the standard you measure everything against. Name a salary in a negotiation and the whole conversation orbits that number, no matter how arbitrary it was. The same thing happens with a life.

The first image of success you absorbed — your parents' version, your neighborhood's version, the version that showed up in every commencement speech you half-listened to — became the anchor. Everything since has been measured against it. Not because you evaluated and chose it. Because it got there first.

You might have updated the details. Swapped the corner office for a startup, the suburbs for the city, the retirement account for a different kind of retirement account. But the underlying shape — the metrics, the timeline, the definition of enough — those are the anchor's. You've been decorating someone else's blueprint and experiencing it as self-expression.

This closes the second escape hatch. "But I really do want this" is worth examining. You might. But the anchor got there before you did, and the brain doesn't distinguish well between a desire it chose and one it inherited.

The familiar became the true.

Psychologists call it cognitive ease. The brain prefers what requires less effort to process, and what you've rehearsed a thousand times requires almost no effort at all. Familiar options feel right. Familiar goals feel chosen. The path you've been on for fifteen years doesn't feel like a default. It feels like a conviction.

This is the mechanism that makes the script invisible. The brain files what's familiar under "just how things are." Questioning it requires a kind of cognitive expense the brain would rather not pay. So it doesn't. And the script keeps running, not because you examined it and agreed, but because examination never seemed necessary.

This closes the third escape hatch. "I've thought about it and I'm sure" may be true. But certainty that comes from familiarity feels identical to certainty that comes from honest evaluation. The brain doesn't flag the difference. It just files both under "mine."

The Question

The Default Script isn't the enemy. It kept you moving when you didn't have a direction of your own. It gave you a framework when you needed one. Every culture writes its own version, and not all of it is wrong.

But a borrowed pattern can't invent. It can only replicate. And at some point, the question stops being whether the life is working and starts being whether the life is yours.

Not what's next on the list. Not what looks right from the outside. Something prior to all of that.

Who first imagined this version of success for you?

Sit with it. Not to produce an answer by Friday. Just to notice what happens when you ask.

Theresa Stroisch is the author of The Imagination Age: Reclaiming Your Most Essential Competence.

Why I Wrote The Imagination Age

People ask me some version of the same question all the time. How did you get from Montana to New York. How did you end up at the Gates Foundation. How do you and Kevin figure out how to live in Southeast Asia for months at a time. The question changes but what they're really asking doesn't: how do you keep doing this?

For a long time I didn't have a good answer. I'd say something about being open to opportunity, or following curiosity, which was true but not useful. It didn't explain the actual mechanism. And I wanted to understand the mechanism — not just for the people asking, but for myself.

So I started researching. I spent months in the neuroscience and behavioral science literature, trying to understand what imagination actually is and how it works. Not imagination as metaphor. Not the word on a motivational poster. Imagination as cognition — something with structure, something the brain does in specific ways. And alongside the research, I kept returning to what I'd seen over twenty years — in boardrooms and team offsites and long dinners with friends across three continents. People who were smart, accomplished, capable — stuck. A leadership team circling the same three options quarter after quarter. A person who built exactly the life they planned and can't explain why it feels hollow. Not because anything was wrong with them. Because something had happened to their thinking and they didn't have a name for it.

What I found is that imagination isn't a personality trait. It's a competence. One that operates in distinct modes, each serving a different function — and each one can be developed or lost. That was also my answer. Every leap I'd made, every unlikely room I'd walked into — it hadn't been courage or luck. It had been imagination doing a specific kind of work.

And then, mid-draft, I watched Demis Hassabis on 60 Minutes. Hassabis is the CEO of Google DeepMind — and a neuroscientist whose doctoral research on memory and imagination was named one of the top ten scientific breakthroughs of the year. He has spent his career at the intersection of how the human brain works and what artificial intelligence can do. And he said, plainly, that imagination is the last thing AI can't replicate. Here was someone who understands both sides of that equation — the neuroscience and the machine — confirming what I was already writing. Imagination is the thing. The human thing.

This is a book about what imagination actually is, how it works, and what happens when we stop using it. If you've been asking how — this is my answer.