April 20, 2026AI and the Future of Work
The genie is not going back in the bottle …
So, let us talk about what actually happens next.
In 1967, a London bank installed one of the first cash machines in the world, and some of the tellers reportedly snuck out and smeared honey on the keyboards. [^1]
Sit with that for a second. Grown adults, professionals, looked at a beige metal box that dispensed cash and were so convinced it had come to take their livelihoods that they tried to sabotage it with breakfast condiments. It is one of the most human things I have ever read, and it is also the entire history of technology anxiety compressed into a single sticky keyboard. People see the new machine, they feel the floor move, and the very first instinct is not "how do I work with this;” it has become "how do I make it stop."
I bring it up because we are doing the exact same thing right now with artificial intelligence, just with worse manners and more LinkedIn posts. And I want to make an argument the evidence supports: AI is not as scary as it has been hyped up to be. The doom is oversold. But — and this is the part nobody wants to hear about. If it is scary, it does not matter, because the genie is out of the bottle. It is not going back in. There is no honey you can put on this keyboard. Which leaves all of us, individuals, and organizations alike, with the oldest and least romantic survival rule there is: adapt or die.
Let me walk through why I believe that, with the receipts.
First, the genie is already out. Surprise.
A lot of the AI conversation is framed as something that is coming, like a storm on the radar. It is not coming. It is here, and it is already in the building.
By the spring of 2026, Gallup found that about half of U.S. employees said they use AI at least a few times a year in their role.[^2] ManpowerGroup's 2026 Global Talent Barometer put regular AI use at 45% of workers — a 13-point jump in a single year.[^3] The World Economic Forum reports that investment in generative AI has grown roughly eightfold since ChatGPT launched, and that 86% of employers expect AI and related technologies to transform their business by 2030.[^4]
Here is the part that should settle the debate about whether this is real: people are using it whether their employers sanctioned it or not. By one 2026 account, around 90% of employees admitted to using personal AI tools for work, while only about 14% of companies were paying for them. [^5] Read that again. Your people did not wait for a memo. They did not wait for a policy. They did not wait for the training. They went and got the tool themselves and started using it for company work, because it made their day easier.
You cannot un-ring that bell. The technology is cheap, it is everywhere, it lives in a browser tab, and the people you employ have already decided it is useful. So, the question was never really "should we let AI into our workplace." That ship sailed, got automated, and is now offering to summarize its own voyage. The only real question is whether the adoption already happening is going to be confident, supported, and pointed at something useful — or scattered, secretive, and slightly panicked. Right now, for most organizations, it is the second one.
So why does everybody feel terrible about it?
Because fear is real, even when it is not rational. And I want to be careful here because it is easy to wave away from other people's anxiety when it is not your job on the line. Trust me – anxiety is something I know VERY well.
The numbers on how people feel are genuinely rough. That same ManpowerGroup survey found that as AI use went up, worker confidence in using technology actually fell by 18% — the first overall confidence decline in three years — and that 43% of workers feared automation could replace their job within two years.[^3] A separate look at the issue found that roughly three-quarters of employees don't feel confident using AI in their daily work, and named the specific 2026 flavor of dread: not just "will I get laid off," but a quieter fear of becoming obsolete — of your skills quietly expiring while you're still using them.[^5] Pew Research has found that more than half of workers worry about AI's impact on their future at work.[^6]
But here is the thing the panic misses. A lot of that fear is not actually about the machine. It is about the silence around the machine.
Look at the gap. One 2026 study found that 92% of executives felt ready to use AI — and only 51% of employees did.[^7] The same research found that just 8% of employees said their company had communicated a clear vision for AI.[^7] So you have a workforce being handed a powerful, unfamiliar tool, told (correctly) that it is going to change their work, and told almost nothing about what that means for them specifically. Of course they are scared. We took the most disorienting workplace shift in a generation and mostly let people fill in the blanks themselves. People do not fill in blanks with optimism. They fill them in with their worst Tuesday-at-3am thoughts.
The fear, in other words, is mostly a communication vacuum wearing a robot costume. And vacuums can be filled. That is good news, not bad.
What happens when a machine takes some of the work?
Let us go back to the tellers, because their story is the most useful thing in this entire conversation — and because it has a twist that keeps everybody honest.
When ATMs rolled out across American banks in big numbers in the 1980s and 1990s, everyone — including some of the bank managers — assumed it was the end of the bank teller. The machine literally had "teller" in its name. And yet, as the economist James Bessen documented, the number of bank tellers did not collapse as ATMs spread. It rose. In 1985 the U.S. had roughly 60,000 ATMs and about 485,000 tellers; by 2002 there were around 352,000 ATMs and about 527,000 tellers. [^8] More machines, more humans. The "robot cashier" the British press feared in 1967 somehow coincided with more people doing the job.
Two things made that happen, and they are exactly the things people forget. First, ATMs made it cheaper to run a branch — the number of tellers needed per branch fell from about 20 to about 13 — so banks opened far more branches to compete, and total teller jobs held up.[^9] Second, and more important: once the machine swallowed the boring part of the job (counting cash, taking deposits), the part that was left got more valuable. Tellers stopped being human calculators and became the relationship — helping customers with the complicated stuff machines could not touch. [^10] The task got automated. The job got upgraded.
Now the honest twist because I refuse to sell you the clean version. The teller job did eventually decline — but not because of the ATM. It started shrinking in the 2010s, and the culprit was the smartphone. Mobile banking meant people simply stopped walking into branches, which cut demand for the relationship work the ATM had protected. [^11] In fact, the World Economic Forum's most recent jobs report now lists bank tellers among the roles expected to keep declining. [^12]
So, which is it — proof that automation does not kill jobs, or proof that it eventually does? Both. Two things are true at once. Automation rarely does what the panic predicts (instant mass unemployment), and it rarely leaves the work untouched forever either. What it reliably does is move the value around. It eats tasks, not jobs, at least at first — and the people who thrive are the ones who follow the value to wherever it went next. The tellers who learned relationship banking were fine for thirty years. The skills that mattered changed from cash-handling to human connection, and the ones who made that move kept working. [^10] That is not a story about robots winning. It is a story about adaptation winning. It always has been.
We have run this experiment repeatedly. The iceman, the elevator operator, the switchboard operator, the travel agent — some jobs really did vanish.[^13] Bookkeeping got eaten by accounting software, and bookkeepers became something closer to advisors.[^10] The lesson across all of it is not "you're safe" and it is not "you're doomed." It is "the work will change under your feet, and your move is to change with it."
The actual numbers are cautionary AND hopeful.
Here is the macro picture, from the most thorough labor study going. The World Economic Forum's Future of Jobs Report 2025, built on surveys of more than 1,000 employers representing over 14 million workers, projects that by 2030 about 92 million jobs will be displaced and about 170 million new ones created — a net gain of roughly 78 million jobs worldwide.[^14] That's churn of around 22% of all jobs in five years.
That is not an apocalypse. It is also not a victory lap. Ninety-two million displaced people are not comforted by a net-positive spreadsheet, and I am not going to insult anyone by pretending the average does not have real human cost underneath it. But the headline that "AI is going to destroy work" is just not what the best data says. The data says work is going to get violently rearranged, with growth in technical and data roles and in deeply human ones like care, teaching, and skilled trades, while clerical, data-entry, and routine administrative roles shrink. [^15]
And then there is the line that should be taped to every desk in the country: the WEF found that nearly 40% of the core skills workers use today are expected to change by 2030.[^14] Notably, that's actually down from a comparable figure of 57% in 2020 — so even the rate of skill churn is itself churning.[^16] The single biggest barrier employers named to navigating all of this wasn't the technology. It was the skills gap, cited by 63% of them, which is why 85% said they plan to prioritize retraining and upskilling their people. [^14]
In plain English: the machines are not the bottleneck. People being ready is the bottleneck. Which, again, is something we can do something about.
The dirty secret: most AI rollouts faceplant — and it is not the tech's fault.
Here is the part that should make everyone breathe a little easier, and it is backed by some genuinely embarrassing statistics.
MIT's 2025 research on AI in business found that around 95% of companies failed to see meaningful return on their AI initiatives within six months.[^17] Boston Consulting Group has put the failure rate of AI projects to deliver expected value at somewhere between 70% and 85% — roughly twice the failure rate of ordinary IT projects.[^17] Gallup's data shows that where AI is helping, the gains are mostly showing up at the level of individual tasks, not yet transforming how whole organizations work.[^2]
Now, why is that good news? Because it means the scary version — the all-powerful AI that walks in and runs the place — is not what is happening on the ground. The tools are capable. The deployments are a mess. And they are a mess for an extremely human reason. Gartner has a name for it: the "enablement illusion," where leaders mistake basic access or a usage metric for actual transformation.[^18] Microsoft researchers describe a "transformation paradox," where companies adopt the tools at speed but never redesign the work around them — and found that only about 26% of AI users said their leadership was even consistently aligned on AI strategy.[^19] The same analysis found that organizational factors like culture, management, and governance accounted for more than twice the variance in whether AI actually delivered, compared to individual skill or mindset.[^19]
Translate that out of consultant-speak and it says something almost reassuring: AI does not win or lose on how smart the model is. It wins or loses on whether the humans around it are led well, trained well, and clear on what they are doing. The advantage does not go to the most technical organization. It goes to the most adaptable one. That is a game where ordinary people and ordinary companies can win.
Managers are the hinge the whole thing swings on
If there is one place to spend your energy, it is the middle. Gartner's early-2026 global survey of more than 12,000 employees and managers found that managers are the ones best positioned to actually integrate AI into daily work — to give people context, encourage experimentation, and make using it feel normal rather than threatening.[^20] The same body of research found that the managers who lead with performance and outcomes were about 20% more likely to hit their goals.[^20]
But we are not setting managers up to do that. Gartner also found that only about 27% of executives reported having a comprehensive AI strategy, and just 20% believed their workforce was truly ready for AI.[^20] We are asking the people in the middle to lead a transformation that the people at the top haven't fully figured out, with teams that haven't been prepared. No wonder it is stalling.
And the fix is not another tool. It's unglamorous human work: a clear vision people can hear, training that connects AI to the specific job in front of someone, managers equipped to lead the conversation instead of dodging it, and the plain acknowledgment that confidence is built, not assigned in a kickoff email. You do not get a workforce to feel capable by telling them the future is exciting. You get there by showing one person, in one role, exactly how this thing makes their actual Tuesday better — and then doing it again, and again.
Adapt or die — and "adapt" is the hopeful word!
So, here is where I land.
Fear is real, but it is mostly fear of the unknown, and the unknown shrinks fast the moment someone turns the lights on. The doom is oversold; the data does not describe a robot apocalypse, it describes a massive, stressful, survivable rearrangement of work — the kind we have lived through before. And the change is permanent. The genie is out. There is no version of this where everyone agrees AI was a bit much and we all go back to 2019.
That leaves adapt or die. I know it is an old, blunt phrase. But I would ask you to hear the hopeful half of it, because most people only hear the threat. "Die" is what happens to the task, the rote step, the thing the machine does better. "Adapt" is what happens to you — and adaptation has a remarkable track record. The teller who learned to sit across from a small-business owner and actually help them kept working for thirty more years. The bookkeeper who became an advisor was fine. The person who smeared honey on the keyboard in 1967 was, I promise you, more afraid that morning than they ever needed to be.
The honey did not stop the machine. It never does. But it also never had to. The job was never really cash-counting. The job was the human part the machine could not reach — and there is, it turns out, an enormous amount of human part left.
We do not get to choose whether the work changes. We only get to choose whether we change with it, on purpose and with our eyes open, or whether we wait, hope it passes, and find out the hard way that it did not. I would rather we choose the first one. And the good news, buried under all the noise, is that it has almost always been a choice we could make.
So go learn the thing. Help your people learn the thing. Lead like humans still matter because the evidence says they are the whole ball game.
And leave the honey for your toast.
Sources
[^1]: Visier, "To Optimize Work Automation, Get Beyond 'Robots Taking Our Jobs'" — recounting the 1967 Barclays cash machine and early teller fears (also cites Bessen's ATM/teller data). https://www.visier.com/blog/optimize-work-automation/
[^2]: Gallup, "Rising AI Adoption Spurs Workforce Changes" (April 2026 survey of 23,717 U.S. employees). https://www.gallup.com/workplace/704225/rising-adoption-spurs-workforce-changes.aspx
[^3]: ManpowerGroup, "2026 Global Talent Barometer" — regular AI usage up 13% to 45%, confidence down 18%, 43% fear automation within two years. https://www.barchart.com/story/news/37128510/
[^4]: Sustainability Magazine summary of the WEF Future of Jobs Report 2025 — eightfold rise in GenAI investment; 86% of employers expect AI to transform their business by 2030. https://sustainabilitymag.com/articles/wef-report-the-impact-of-ai-driving-170m-new-jobs-by-2030
[^5]: People Managing People, "Employee AI Fears in 2026: What Actually Kills Adoption" — ~75% lack confidence; ~90% use personal AI tools while ~14% of firms pay; the "fear of becoming obsolete." https://peoplemanagingpeople.com/workforce-management/ai-fears-2026/
[^6]: Pew Research, cited in People Managing People (above) 52% of workers worry about AI's impact on their future at work.
[^7]: BetterWorks, "2026 State of Performance Enablement Report" (via BusinessWire) vs. 51% of employees feel ready; only 8% say their company communicated a clear AI vision. https://www.businesswire.com/news/home/20260303528573/en
[^8]: Visier (citing James Bessen, Learning by Doing) — 1985: ~60,000 ATMs / ~485,000 tellers; 2002: ~352,000 ATMs / ~527,000 tellers. https://www.visier.com/blog/optimize-work-automation/
[^9]: James Bessen, "Toil and Technology," IMF Finance & Development (March 2015) — tellers per branch fell from ~20 to ~13 (1988–2004) while urban branches grew ~43%. https://www.imf.org/external/pubs/ft/fandd/2015/03/bessen.htm
[^10]: ATM Marketplace, "Tellers, toil and technology" (drawing on Bessen) the shift from cash-handling to relationship banking; accounting software and bookkeeping. https://www.atmmarketplace.com/articles/tellers-technology-and-atms/
[^11]: CCIA, "What Bank Tellers and Radiologists Can Tell Us about Our Job Security in the AI Era" (Feb 2026) teller decline in the 2010s driven by mobile banking, not ATMs. https://ccianet.org/articles/what-bank-tellers-and-radiologists-can-tell-us-about-our-job-security-in-the-ai-era/
[^12]: People Matters / WEF Future of Jobs Report 2025 — bank tellers listed among declining roles. https://sea.peoplemattersglobal.com/article/employee-skilling/170-million-jobs-to-emerge-by-2030-but-92-million-at-risk-wef-warns-48552
[^13]: Brookings, "Not all robots take your job, some become your co-worker” historically eliminated roles (iceman, elevator operator, travel agent) vs. transformed ones. https://www.brookings.edu/articles/not-all-robots-take-your-job-some-become-your-co-worker/
[^14]: World Economic Forum, Future of Jobs Report 2025 (press release) 170M jobs created, 92M displaced, net +78M by 2030; 22% churn; ~40% of core skills change; 63% cite skills gap as top barrier; 85% to prioritize upskilling. https://www.weforum.org/press/2025/01/future-of-jobs-report-2025-78-million-new-job-opportunities-by-2030-but-urgent-upskilling-needed-to-prepare-workforces/
[^15]: World Economic Forum, "Future of Jobs Report 2025: the jobs of the future” growth in tech/data and in frontline/human roles (care, teaching, trades); decline in clerical and administrative roles. https://www.weforum.org/stories/2025/01/future-of-jobs-report-2025-jobs-of-the-future-and-the-skills-you-need-to-get-them/
[^16]: Sustainability Magazine (WEF data) skills-disruption figure of 39% by 2030, down from 57% in 2020. https://sustainabilitymag.com/articles/wef-report-the-impact-of-ai-driving-170m-new-jobs-by-2030
[^17]: MIT State of AI in Business 2025 and BCG research, summarized in "Why Do AI Implementations Fail Without Change Management?" — ~95% see no meaningful ROI within six months; 70–85% of AI projects fail to deliver expected value. https://aismartventures.com/posts/why-do-ai-implementations-fail-without-change-management-the-people-side-of-ai-transformation/
[^18]: Gartner, "By 2027, 50% of Enterprises Without a People-Centric AI Strategy Will Lose Their Top AI Talent" — the "enablement illusion." https://www.gartner.com/en/newsroom/press-releases/2026-05-13-gartner-predicts-by-2027-50-percent-of-enterprises-without-a-people-centric-ai-strategy-will-lose-their-top-ai-talent
[^19]: Gloat, "AI Workforce Trends 2026” Microsoft's "Transformation Paradox"; only 26% of AI users say leadership is consistently aligned; organizational factors account for 2x+ the variance in AI impact. https://gloat.com/blog/ai-workforce-trends/
[^20]: Gartner Global Labor Market Survey, Q1 2026 (12,004 employees/managers, 40 countries) — managers best positioned to integrate AI; performance-first managers ~20% more likely to meet goals; only 27% of executives have a comprehensive AI strategy and 20% believe their workforce is AI-ready. https://www.gartner.com/en/newsroom/press-releases/2026-05-13-gartner-predicts-by-2027-50-percent-of-enterprises-without-a-people-centric-ai-strategy-will-lose-their-top-ai-talent
