People argue about whether Mark Twain said this, but whether he did or not it’s a great quote and one that feels particularly relevant when a new technology makes us believe that everything has changed. This has happened thousands of times before – the horse, the car, electricity – all of these changed the world by giving us a power we didn’t have before. Right now that technology is artificial intelligence.
AI tools are amazing – they’ve become scarily capable and they get better every day. They do in seconds things that used to take hours or days and we’re all excited about the new powers they give us. Everyone is experimenting – they’re vibe-coding, designing sites, writing briefs, producing campaigns. They’re doing things that previously needed experienced professional support and some of the results are genuinely impressive. But some (possibly most) are not.
In these times of technological excitement it’s easy to forget that having access to powerful tools is not the same thing as having the skill to use them well.
There’s more to being a carpenter than owning a saw.
We’ve been here before
Those of us who are old enough to remember the 80s are probably detecting strong echoes of an earlier technological revolution – the arrival of desktop publishing.
Before DTP, the job of producing professional printed materials was firmly in the hands of designers, typesetters, artworkers, printers and repro houses. It needed specialist knowledge, expensive equipment and a complex production process. Producing a brochure, newsletter or advert wasn’t something most people could easily do from their own desk.
Then came the Macintosh, affordable laser printers, page layout software and tools like Quark Xpress and Aldus PageMaker (the great grandaddy of Adobe Indesign). Suddenly, ordinary office users could create newsletters, flyers, posters and leaflets themselves. The barriers to entry fell dramatically and tools that had once been part of the arcane skill-set of trained specialists became available to almost anyone with a mouse and a desire to use it. For those of us who saw it happen, it was a remarkable shift. It was also a visual disaster.
The democratisation of design didn’t instantly create a generation of designers. It created a generation of people with access to design software. Those are not the same thing.
For the first time, huge numbers of people could choose from dozens of fonts, stretch text, add drop shadows, place clip art, create borders, add page curls. And because they could, they often did. All at once. On the same page.
The result was a flood of newsletters and brochures full of visual noise: too many typefaces, inconsistent spacing, poor hierarchy, awkward alignment, low-quality images and layouts that were technically possible but visually exhausting.
The problem wasn’t the technology – it was the assumption that access to the tool replaced the need for taste, judgement and skill. It didn’t. Believe me, it didn’t.
And to nobody’s surprise, professional designers didn’t become redundant because everyone could now open a page layout programme. If anything, the difference between amateur and professional work became more obvious. The tools made production easier, but they didn’t remove the need to understand communication, proportion, typography, structure, audience, readability or restraint. The same thing is happening now with AI.
AI has lowered the barrier to production
AI has not just made certain tasks easier. It has made them feel deceptively easy. This is a subtle but important distinction.
A business owner can now ask their favourite AI to write a blog post, generate a headline, crack out a set of wireframes or produce a hundred social media captions. Within seconds, something appears. It is formatted, it sounds plausible and may even be grammatically polished.
Compared with a blank page, this feels miraculous and in many ways, it is. AI can be an excellent assistant. It can help you get unstuck. It can speed up repetitive work. It can offer alternative phrasings, summarise complex material, turn rough notes into usable copy and help non-specialists express ideas more clearly. Used well, it can be genuinely useful. But the phrase “used well” is doing a lot of work here because AI doesn’t remove the need for judgement – it dramatically increases the importance of judgement.
When a tool can produce large volumes of content almost instantly, the limiting factor is no longer production. The limiting factor is selection, direction and quality control. What should we ask for? What is good? What is generic? What is inaccurate? What is off-brand? What is persuasive? What’s appropriate for this audience, on this page, at this stage of the customer journey? AI can generate options but it can’t magically know which option is right. That is where skill still matters.
Output is not the same as outcome
One of the traps with new technology is that we become impressed by the wrong thing.
Sure, DTP made it possible for people to produce printed pages but the goal was never simply to put words on a page – it was to communicate clearly, persuade an audience, sell a product, explain an idea or build trust.
AI makes it possible to produce words, images, plans, code and layouts but again, the goal isn’t to simply to produce stuff. The goal is to produce the right stuff.
- A website isn’t successful because it contains words. It is successful because those words help the right people understand the right message and take the right action.
- A brand isn’t strong because it has a logo, a colour palette and some social posts. It is strong because those elements work together to create recognition, consistency and trust.
- A blog post isn’t valuable because it is 2,000 words long. It is valuable because it says something worth reading, in a way that is useful to the reader and credible for the organisation publishing it.
AI can help with all of these things, but it doesn’t automatically solve your problems – this is why AI-generated work can vary so wildly. We’ve seen customers produce great results with AI tools: clear first drafts, useful research summaries, thoughtful content outlines and practical starting points for discussion. But we’ve seen as much or more that is bland, confused, inaccurate, overcomplicated or simply wrong for the audience. This isn’t a difference in technology – it isn’t a Claude vs ChatGPT conversation. The difference is rarely the tool itself. It is the person using it.
Someone who knows what they’re doing can guide the tool, challenge it, refine it and recognise when it has gone wrong. Someone who doesn’t know what they’re doing will probably take the first answer because it looks professional enough. That’s dangerous, because AI is very good at sounding confident.
The illusion of competence
One of the reasons AI is so powerful is also one of the reasons it is so risky: the stuff it produces looks polished.
A poor layout created in early desktop publishing software often looked obviously amateurish. The fonts clashed. The spacing was strange. It was full of crap low-res clipart. AI-generated content can be harder to judge because it often looks competent on the surface. The sentences flow. The structure seems sensible. The tone may be smooth and professional. There may be headings, bullet points and a neat conclusion.
But polish isn’t the same as quality. A piece of AI-generated copy can be grammatically correct and strategically useless. It can be fluent but generic – confident but inaccurate. It can say a great deal while communicating very little. This is the new version of the desktop publishing problem.
In the 1980s, people mistook access to layout tools for design ability. Today, people risk mistaking access to AI tools for creative, strategic or technical expertise. The tool can imitate the surface of professional work but it can’t guarantee the substance.
Good prompts help, but they are not the whole answer
The internet is full of advice (ironically, often AI-generated) about prompt writing, and some of it is useful. Better instructions usually produce better results. Clearer context helps. Examples help. Constraints help. Iteration helps. But don’t be fooled – prompting isn’t magic.
A good prompt can improve the output, but it can’t replace the knowledge needed to judge that output. If you ask AI to produce website copy, you still need to understand your audience, your offer, your positioning, your tone of voice and the action you want users to take. If you ask it to suggest a website structure, you still need to understand user journeys, content hierarchy, accessibility, search visibility and maintainability. If you ask it to generate design ideas, you still need to understand brand fit, visual hierarchy, usability and implementation. If you don’t, you’re just delegating decisions you’re ill-equipped to evaluate.
This is where AI can create a false sense of security. It gives people something that feels finished, but something that feels finished isn’t the same as something that is finished. In professional work, the first draft is rarely the final answer – it’s the start of something, not the end of anything. Experienced pros know what to keep, what to remove, what to question and what to improve. That’s the skill.
Tools change the work, not the need for expertise
None of this means you should dismiss or avoid AI. Quite the opposite.
Desktop publishing didn’t destroy design – It changed it. It changed workflows, expectations, timescales and costs. It removed some specialist production tasks but created new ones and it made visual communication more accessible. Professional designers adopted the tools too, and used them to do better work faster.
AI is doing the same, across a huge range of fields. Without doubt, it’ll change how websites are planned, written, designed, built and maintained. It’ll make some tasks quicker. It’ll make some services more efficient. It’ll help clients participate in the creative process. It’ll reduce the friction involved in getting ideas out of people’s heads and onto the page. Obviously, this is all good, but it won’t remove the need for expertise. Instead, it’ll change where that expertise is needed.
The weight of effort will move away from producing every word, image or function from scratch towards asking better questions, shaping better structures, editing more intelligently, checking accuracy, maintaining consistency and making sound decisions. In other words, the skill moves up the chain.
The professional isn’t valuable just because they’ve got tools the client does not – increasingly, clients will have access to many of the same tools. The professional is valuable because they know what good looks like. They understand the consequences of choices. They can see the gap between something that’s just acceptable and something that is genuinely effective. This distinction matters.
The amateur asks “can I make this?” The professional asks “should I?”
New tools encourage experimentation, and experimentation is healthy. But there is a difference between possibility and judgement.
- AI can generate ten homepage headlines. Should any of them be used?
- AI can write a 3,000-word article. Does the audience need it?
- AI can create a set of illustrations. Do they support the brand?
- AI can suggest features for a website. Are they useful, affordable and maintainable?
- AI can turn a simple brief into a very long and detailed specification. Has it made the project clearer, or has it accidentally made it more complicated and expensive?
That last point is particularly important. We are already seeing AI used to generate briefs, specifications and requirements documents. Sometimes this is helpful. A client with rough thoughts can use AI to organise them into a clearer form. But AI has a tendency to add detail. It can over-specify. It can include features, integrations, assumptions and edge cases that sound sensible in isolation but may be unnecessary for the project. A simple website brief can quickly become a sprawling document full of requirements the client does not really need, does not fully understand and may not have the budget to implement.
Again, the issue is not the tool. The issue is the absence of experienced judgement. AI can help write a brief. But someone still needs to know whether the brief is any good.
Craft is partly knowing what to leave out
One of the most obvious signs of professional work is restraint. A good designer doesn’t go font-tastic. A good developer doesn’t install every possible plugin. A good copywriter doesn’t include every argument. A good carpenter does not use every tool in the box just because it is there. Much of the art of professional judgement is in knowing what to leave out.
This is where AI can be both useful and risky. It’s great at generating more: more ideas, more text, more options, more variations. But more is not always better. In many projects, the real value comes from reduction: simplifying the message, clarifying the structure, removing distractions and focusing attention. AI can help create the raw material. It is less reliable at knowing when enough is enough.
That’s why human taste still matters. Taste isn’t just decoration – it is the ability to choose, to recognise proportion, tone, relevance and fit. It’s the ability to see that something may be technically correct but still wrong.
In website work, this comes up constantly. A page can contain all the right info and still fail because the hierarchy is weak. A design can include all the brand colours and still feel wrong. A feature can work technically and still make the user experience worse. A piece of copy can be accurate and still fail to persuade. You need to apply human knowledge and human understanding to your work – you can’t rely on your friendly digital assistant to do your thinking for you.
AI is a collaborator, not a substitute for thinking
The healthiest way to approach AI is not as a replacement for skill, but as a collaborator that amplifies skill.
- For someone who knows what they are doing, AI can be extremely powerful. It can speed up early drafts, explore alternatives, identify gaps and handle repetitive tasks. It can make good people faster and, in some cases, better.
- For someone without the underlying skill, AI can still be useful, but the risks are greater. It may produce work that looks more advanced than it is. It may hide weak thinking behind polished language. It may encourage people to publish, send or build things before they have been properly considered.
This isn’t unique to AI – it is true of most powerful tools.A camera doesn’t make you a photographer. A spreadsheet doesn’t make you an accountant. A page layout programme doesn’t make you a designer. A saw doesn’t make you a carpenter. The tool matters. But skill matters more.
What this means for businesses
For businesses, the practical lesson is not to avoid AI. The lesson is to use AI thoughtfully.
Use it to get started – to explore, to speed up repetitive work, to turn messy thoughts into something you can review. Use it to create first drafts, not final answers. Use it to support people who already understand the problem they are trying to solve. Just be careful about outsourcing judgement.
Before using AI-generated work, ask some basic questions. Does this reflect who we are? Is it true? Is it useful? Is it specific enough? Is it appropriate for our audience? Does it support our goals? Is it saying anything distinctive? Has it added complexity we don’t need? Would we be happy to stand behind it? If you’re not sure, bring in the right expertise.
That may be a designer, developer, copywriter, strategist, marketer or subject specialist. Not because those people have access to secret tools, but because they bring experience, judgement and taste to the process.
In the long run, AI may make that expertise more valuable, not less. When anyone can produce something, the ability to produce the right thing becomes more important.
The rhyme of history
The desktop publishing revolution gave people unprecedented creative power. It changed publishing forever. It democratised production, speeded up production and lowered costs. But it also taught an important lesson: making the tool available does not automatically transfer the craft.
AI is teaching us the same lesson again. The technology is different, the scale is different, the speed is different, but the rhyme is familiar:
- A new tool arrives.
- It lowers the barrier to entry.
- People experiment.
- Some results are brilliant. Some are terrible.
- For a while, there is confusion between access and expertise.
- Then, gradually, we remember that the tool is only part of the story – the real value lies in how it is used.
AI will continue to change the way we work. It’ll become part of everyday business, creative and technical processes. It’ll help people do things they couldn’t previously do, more quickly and more cheaply but it won’t remove the need for skill.
Because there is more to being a designer than owning design software. There is more to being a developer than generating code. There is more to being a writer than producing words. There is more to being a strategist than asking for a plan.
And there is still, very definitely, more to being a carpenter than owning a saw.
