Before the modern Content Management System (CMS), publishing to the web was a technical endeavour.
Many early websites were built with hand-crafted HTML, and while tools such as MS Frontpage and Dreamweaver, and database-driven platforms that included PHP and Classic ASP front ends later reduced the need to code every single page, you still required a significant degree of technical knowledge to publish.
Then, CMS, and later Enterprise CMS, revolutionised everything. They truly opened up content creation, allowing anyone with basic IT skills to edit and add content to a website.
And over the following years we saw the consequence: a proliferation of truly awful websites.
In some cases, these were vast digital estates containing tens of thousands of pages, authored by hundreds of people with little to no understanding of the disciplines of content design.
In a previous role I was given the remit of managing that organisation's website with a view to improving its content and functionality.
After a review, my team conceded that the content, published by professionals in their field through a well-intentioned but devolved model using an enterprise CMS, was simply beyond salvaging, and the most effective way to improve the website was not to review and rewrite the myriad of content, but to start again.
Ten years later I saw this pattern again.
Low-code form builders transformed the creation of digital services. As with web pages, no longer do you need a developer to hand-code a form. By simply dragging and dropping elements, a functional form can be built in minutes.
But, much like the CMS revolution, this empowerment enabled some organisations to quickly and easily create some truly unusable forms, often built by people with little grasp of user experience (UX) or digital accessibility.
And now, I see the potential for history to repeat itself again with application development.
The emergence of “vibe coding” through generative AI platforms allows the creation of software simply by describing how it should work. The code is then generated from a prompt and the parallels to my previous examples are striking.
Just as effective content requires an understanding of style and grammar, and good forms require knowledge of usability and accessibility, creating quality software demands a foundational knowledge of secure, scalable architecture and robust engineering practices.
This is not to say that AI isn’t a profoundly useful tool in development, it is. Initiatives like a new Government Digital Service to promote the use of AI coding assistants will undoubtedly save developers immense amounts of time. The crucial distinction, however, is using AI as an assistant to assist the professional, not as a replacement for the expertise they bring.
The lesson remains the same: empowerment without governance and expertise often leads to poor outcomes, just ask your content and service designers, and the principle endures: just because advancements in technology mean you can, doesn't mean you should.
Comments
Post a Comment