Artificial intelligence is the most discussed technology of the moment. Barely a week passes without a new announcement, a new capability, a new warning. And in organisations across every sector, people are grappling with a familiar feeling, the mixture of curiosity, anxiety and uncertainty that comes whenever something genuinely new arrives at the door.
Some people worry their role will disappear. Some worry that colleagues are using AI to pass off work that isn't really theirs, making it impossible to judge genuine capability. Some leaders feel pressure to have a strategy before they really understand the thing they're supposed to be strategising about. Most people, if they're honest, are just trying to work out what it means for them.
These are all legitimate responses. They deserve a serious answer, not reassurance, not hype, but genuine perspective. And the best perspective available is history.
We Have Been Here Before
In the early 1800s, a movement of skilled textile workers in England began breaking into factories and destroying the mechanical looms that were displacing them. They called themselves Luddites, after a possibly mythical figure named Ned Ludd who had smashed two stocking frames in a fit of rage years earlier. The British government treated the uprising so seriously that at one point there were more troops deployed against the Luddites than Napoleon had facing him in the Peninsular War.
History has not been kind to the Luddites as a label, to call someone a Luddite today is to call them backwards. But the actual Luddites were not stupid people who failed to understand progress. They were skilled craftsmen who understood, very precisely, that a specific technology was going to eliminate the value of skills they had spent years building. They were right about that. The question they couldn't answer, because nobody could at the time, was what would come next.
What came next was more jobs, different jobs, and eventually a standard of living that the weavers of 1811 could not have imagined. But that truth was of no comfort to the people living through the transition, and it would be glib to pretend otherwise.
The printing press, when Gutenberg introduced moveable type in the 1440s, was viewed with deep suspicion by the Catholic Church and the scribal class who had controlled the production and interpretation of knowledge for centuries. Manuscripts were things of beauty, copied with skill and care, each one a small act of devotion. The press was mechanical, impersonal, and terrifyingly fast. More fundamentally, it threatened the monopoly on meaning. Who would control what people read, and therefore what they believed, if anyone with a press could print anything?
The Church's fears were not irrational, the Reformation, the scientific revolution and the Enlightenment all followed in the press's wake. The world that emerged was unrecognisable from the one the scribes had known. But it was also vastly richer in ideas, voices and human possibility.
When electricity became available to manufacturers in the 1880s, something curious happened. Factory owners, rational, commercially driven people, largely failed to benefit from it for nearly thirty years. The reason was that they installed electric motors the same way they had used steam: one large engine in the centre of the factory floor, with leather belts and shafts distributing power to machines arranged around it. They replaced the energy source but kept the architecture.
The productivity gains only came when a new generation of managers, many of whom had grown up with electricity and therefore didn't think of it as a substitute for steam but as a fundamentally different thing, redesigned their factories entirely. Individual motors on each machine. Flexible layouts. Workers who could think about tasks rather than managing the physical transmission of power. Output per worker roughly doubled. But it took thirty years and a generational shift in thinking to get there.
When ATMs arrived in banks during the 1970s, the prediction was straightforward: machines would replace tellers, bank branches would close, jobs would vanish. The opposite happened. Because ATMs made it cheaper to operate a bank branch, banks opened more of them. Because tellers were freed from routine cash transactions, they moved into higher-value work, advising customers, selling products, managing relationships. The number of bank teller jobs in the United States actually increased in the decades after the ATM's introduction. The nature of the job changed profoundly. The number of jobs did not fall.
The spreadsheet tells a similar story. When VisiCalc and then Lotus 1-2-3 arrived in accounting departments in the early 1980s, the fear was that they would eliminate the need for accountants. In reality, they eliminated the need for rooms full of people doing arithmetic, and elevated the people who remained into genuine analysis. The question shifted from "what are the numbers?" to "what do the numbers mean?" Demand for accountants rose. The profession became more intellectually demanding, more strategic and better paid.
What the Pattern Actually Shows
Look across these examples and a pattern emerges that is more useful than simple reassurance. In almost every case:
The fear was real and the disruption was real. The Luddite weavers were not wrong that their specific skills were being devalued. The scribes were not wrong that their role was disappearing. The tellers were not wrong that cash handling was being automated. The disruption was genuine.
But the organisations, and people, who fell furthest behind were rarely the ones who refused to adopt the technology. They were the ones who adopted the technology without changing how they worked. The factory owners who put one big electric motor where the steam engine had been and called it done. The accounting firms that used spreadsheets to do faster arithmetic rather than asking different questions. They got some efficiency gains. They missed the transformation.
And the most dangerous position of all was to feel that because you had adopted the technology, you had dealt with it.
Why AI Is Genuinely Different, and Why That Matters
Most previous technological shifts automated specific, defined tasks. Looms wove cloth. ATMs dispensed cash. Spreadsheets calculated numbers. Each was powerful, each displaced specific work, but each was bounded.
Large language models are different in kind, not just degree. They don't automate a task, they automate a cognitive capability: the ability to read, synthesise, draft, reason, translate and generate across an enormous range of domains. That's a different sort of thing. It is closer to a general-purpose tool than anything we have introduced into organisations before, and general-purpose tools reshape work in ways that are harder to predict and harder to manage.
This is why organisations that are treating AI adoption as an IT deployment are making the same mistake as the factory owners with the electric motor. They are changing the energy source and keeping the architecture. They will get some efficiency gains. They will miss the transformation.
It is also why the anxiety people feel is not silly or backward. It is an appropriate response to genuine uncertainty. The question is not whether AI will change how knowledge work is done, it will, but what that means for your organisation, your people, and the capabilities that will matter in the world it creates.
How to Actually Introduce AI Into an Organisation
The first thing to accept is that this is a change programme, not a technology deployment. The distinction matters. A technology deployment has a go-live date and a training plan. A change programme has a destination, a route and honest acknowledgement of what it asks of people.
Start with the work, not the tool. Before asking which AI tools to buy, map where knowledge work actually happens in your organisation, where people are reading, synthesising, drafting, analysing, advising. That is where the leverage is. The tools should follow the work, not the other way around.
Name the anxiety and take it seriously. People who are worried about AI taking their jobs are not being irrational. They deserve a genuine response, not a corporate communication about exciting opportunities. The honest answer for most organisations is: some roles will change significantly, new capabilities will be needed, and we don't yet know exactly what that looks like. Saying that clearly, with a commitment to navigate it together, is more trustworthy than any amount of reassurance.
Distinguish between automation and augmentation, and be honest about which you're doing. Using AI to reduce headcount is a legitimate business decision. Using AI to make your existing people dramatically more capable is a different decision. Both can be right in different contexts. But conflating them, talking about augmentation while quietly planning automation, destroys trust faster than almost anything else.
Invest in the learning, not just the licence. The capability gap in AI adoption is not primarily about access, it is about understanding. Most people have the tool but not the mental model for using it well. Prompt engineering, understanding what LLMs are actually doing, knowing where they are unreliable, these are skills that require real investment to develop, not a one-hour onboarding session.
Build communities of practice, not just training programmes. The people in your organisation who are getting remarkable results from AI are probably not the ones who attended the most training. They are the ones who are experimenting, sharing what they find, and building shared understanding of what works. Create the conditions for that to happen deliberately.
Change the processes, not just the tools. Going back to the factory floor: the transformation came when people redesigned the work around the new capability, not when they used the new capability to do the old work faster. Ask what becomes possible that wasn't possible before. What questions could you now answer that you couldn't? What turnaround times could change? What could one person now do that previously required a team? Then redesign backwards from those possibilities.
Take governance seriously without using it as a reason not to start. There are genuine questions about data security, intellectual property, accuracy and bias in AI-generated content. These need answers, not dismissal. But they are answerable questions, and organisations that use them as reasons to do nothing will simply fall behind while other organisations answer them. Proportionate, practical governance, clear on where AI can and cannot be used, with accountability for outputs, is achievable and necessary.
The Productivity Opportunity Is Real
It is worth being direct about what is at stake on the upside, because it is easy for the complexity of the change conversation to crowd out the scale of the opportunity.
Organisations that use large language models well, genuinely well, not just for summarising meeting notes, are seeing knowledge workers operating at a level that was previously impossible. Analysis that took a week done in a day. First drafts of complex documents produced in minutes and refined by human expertise rather than built from scratch. Regulatory and compliance review that previously required specialist input completed as a first pass by a generalist with a well-configured tool. Research that used to mean weeks in a library compressed into hours.
The compounding effect of this across an organisation, if the capability is genuinely embedded rather than bolted on, is transformational. Not in the sense that word is usually deployed, which is as an aspiration. In the sense of genuinely changing what the organisation is capable of producing, with the people it already has.
The Luddites were skilled, intelligent people who saw clearly that something was changing and whose response was to try to hold the line. The factory owners who outcompeted them were not necessarily smarter or more visionary, many of them also failed to exploit electricity properly for thirty years. The ones who succeeded were the ones who combined honest acknowledgement of what was changing with genuine willingness to redesign how work was done.
That combination, honest about the disruption, serious about the change, clear-eyed about the opportunity, is what good AI adoption looks like. It is not a technology question. It is a change question. And change, done properly, is something organisations can genuinely navigate.
