The Curse of Accessibility

When I first came across the news that PMX would have an AI version of himself capable of speaking and interacting with the public, I wasn’t amazed by the technology. Far from it.

Rather, I was surprised by how unsurprising it felt.

Half a decade ago, the idea of a sitting prime minister having a digital double would’ve sounded like something out of a (dystopian) science fiction film. Today, it barely registers beyond just another headline in our feeds. We acknowledge it, maybe share it with a friend as a joke, and continue doomscrolling.

That was the moment I realised there seems to be a slow and seismic shift. It wasn’t just that AI had become more capable. It had become ordinary, the norm.

And once I noticed it, it’s really hard to unsee it.

Let’s start with government agencies publishing official announcements with unmistakably AI-generated illustrations. You can spot them from a mile away: the same glossy, hyper-real aesthetic, the same uncanny expressions, the same generic typography awkwardly stitched into the composition.

What irks me isn’t merely that they’re AI-generated, but how bland and interchangeable they all feel. Posters from entirely different agencies begin to look as though they came from the same prompt rather than from organisations with their own visual identity and design language.

If that’s not enough, there’s TV3 replacing its long-standing adhan visuals with AI-generated footage, much to everyone’s chagrin. The newly appointed management of my condominium has also started accompanying residents’ announcements in our Telegram group with AI-generated posters.

Don’t get me started with the amount of AI-regurgitated content I frequently stumbled upon on LinkedIn. Heck, even the mamak I regularly visit has replaced photographs of its dishes with AI-generated renderings on its digital menu.

None of these examples are particularly offensive in and of themselves. If anything, they’re understandable from a cost standpoint. One cannot deny that using a free AI model is significantly cheaper than commissioning an illustrator, faster than organising a photoshoot, and infinitely more convenient than waiting for revisions.

Somehow, AI has become the obvious answer. The next big thing. The change that everyone is somehow FOMO-ing about. And the more I thought about it, the more I realised that AI capabilities themselves isn’t what unsettles me.

Accessibility is.

---

One of the things I’ve always loved about software or technology in general is how relentlessly it removes barriers.

Frameworks abstract away boilerplate so we can focus on solving the problems themselves instead of wiring everything from scratch. Design systems mean we no longer have to reinvent the same button for every project. Every new tool seems to shave off another layer of friction, giving us more time to focus on what actually matters.

Naturally, AI is simply the latest step in that progression. It lowers the barrier to entry even further, making it possible to generate code, images, copy, and even videos in a matter of seconds.

As a frontend engineer, I should be celebrating that. And in many ways, I do.

Compared to just two years ago, AI has become an integral part of my workflow. On any given workday, you’ll find me with Claude Code open alongside Figma’s MCP, bouncing between prompts and code as though they’re part of the same language. In fact, the days are more common where I would genuinely spend more time writing prompts than I do writing code. I’ve even caught myself tidying up our Jira board, attending to other menial tasks and frankly scrambling for work to keep myself occupied while waiting for my usage limit to reset.

What usually took me days to develop the same UI, I can do so with ease in just a few hours with extra time to doomscroll in between. My product manager now could just easily visualize her PRDs into high-fidelity Figma mockups without the help of UI/UX designers if she chooses to. It’s difficult to overstate just how transformative these tools have been. They haven’t replaced the work but they’ve changed the very shape of it.

Over the past year however, I’ve started noticing something that feels… different.

It isn’t the technology itself that bothers me. It’s how effortlessly and quietly it has become the default answer, to the point where almost nobody seems to question whether they should.

As designers and engineers, we’ve spent decades trying to make creation more accessible. Yet AI exposed something I hadn’t really considered before.

Accessibility doesn’t just lowers the cost of good work. It lowers the threshold for unintentional work.

When almost anyone can produce something that looks finished, effort is no longer a reliable proxy for thought. When creating something requires almost no effort, it also requires very little intention. Intention matters a lot in producing quality, meaningful work.

In many ways, AI has also blurred the relationship between expertise and its visible expression.

For a long time, a thoughtfully designed interface, a well-composed photograph, or a beautifully illustrated poster served as evidence of the craft behind it. Today, those same visual signals can often be reproduced with a prompt. The appearance of expertise has become dramatically more accessible than the expertise itself—and that’s precisely why intention matters.

Good design has never been about producing something that looks polished. If anything, polish is often the least interesting part of the process.

Good design is the accumulation of countless small decisions. Not just what something should look like, but why it should exist in the first place.

Why should this information appear before that one?
Why should this interaction feel immediate?
Does this notification deserve to interrupt the user, or can it wait?
Does this colour communicate trust or urgency? Should this even be a button at all?

Users rarely notice those decisions individually, but they experience the result collectively. The interface is merely the final output of hundreds of conversations and interations that happened long before anyone opened Figma.

AI is remarkably good at reproducing the appearance of that outcome. It can imitate aesthetics surprisingly well—emphasis on the word ‘imitate’. What it cannot do is replace the thinking that gave those aesthetics purpose in the first place.

Perhaps that’s why so much AI-generated content feels oddly hollow, soulless even. It isn’t necessarily ugly, nor is it technically incorrect. It simply feels like something that exists because it could, rather than because someone made a deliberate decision that it should.

This is the main trade-off of striving to make creation almost completely frictionless. For years, we’ve treated friction as the number one enemy. We automate repetitive tasks, optimise workflows, and eliminate unnecessary steps because those things help people focus on meaningful work. Sometimes we even do so excessively and prematurely.

But some forms of friction serve an important purpose. They force us to pause, reconsider, and justify our choices. When generating a poster takes merely a few seconds instead of three days, there’s far less incentive to ask whether the poster communicates the right message, whether it represents the organisation well and its brand identity, or whether another approach might have been more appropriate.

The conversation quietly shifts from:

“What should we make?”

to:

“Can we generate something that’s good enough?”

Those are two fundamentally different questions.

---

I don’t think AI is diminishing creativity. If anything, it has empowered people who previously lacked the technical skills to express their ideas—and that’s undeniably valuable. I’d be remiss not to admit to using it to polish my writing ideas and proses myself, including this one.

What worries me is how easily convenience begins to replace intentionality. Once every solution is only a prompt away, choosing not to generate something starts to feel like wasted effort. Creation slowly becomes a matter of convenience rather than an act of care. Less an extension of the people behind it, and more a product of whatever the model returns.

And in chasing convenience, we quietly relinquish something far more valuable: the opportunity to wrestle with our ideas, to make the small decisions that give our work its character, and to leave behind something that we can genuinely and proudly call our own. Slowly, we’re unconsciously surrendering the very process that gives our work meaning.

In prioritising convenience above all else, we risk forfeiting our right to be deliberate creators.

I suppose that’s the real double-edged sword of accessibility. Not that AI makes creation accessible to everyone, but that it quietly erodes the moments of discernment that once shaped what we chose to create.

Good software doesn’t happen by accident. Neither does good writing. Neither does good design.

They’re the result of countless small, deliberate decisions that most people will never notice, but almost everyone will experience. AI can generate the artefact remarkably well. What it still cannot generate is the care, judgement, and intention behind those decisions.

May we never lose our willingness to wrestle with our ideas, for it is in that struggle that our work becomes unmistakably our own.