So Are We Bound to Be Replaced by AI?

My supervisor once said — back when ChatGPT had just appeared and before every company felt compelled to convince the world that they were an ‘AI company’ for fear of seeming old and outdated — ‘I’m not sure whether AI will replace humans, but AI will certainly replace people who don’t know how to use it.’

As sharp as that line is, it touches a deep fear (or excitement?) many of us share: is this the end of humanity as we know it? As AI becomes increasingly powerful, will we suddenly find ourselves out of work? Will we one day live in a world where androids (no pun intended) do everything for us while we simply sit at home being served?

What Can AI Replace, and What Can’t It?

Think about it: what do you actually want AI to replace?

Do you want a tool that creates the perfect product without your writing a single line of code? Something that translates foreign articles into something you can genuinely read? When you look closely, you’ll notice that these are all chores.

We enjoy the creative part of building things, but not the tedious labour that comes before it.

We like talking to people around the world, but don’t necessarily enjoy grinding through the process of learning their language.

If you reflect on it, you’ll realise something important:

We only ask AI to do the things we don’t want to do.

You would never ask an AI chatbot to browse X, Threads or TikTok for you. The joy of those platforms lies in the interaction — perhaps even the chemistry — between you and the content, not in maximising your time by having someone (or something) summarise it for you.

AIs Are Fast

AI models can generate large amounts of content in seconds — far faster than any human, though still slower than a well-optimised traditional programme. But what they gain in output speed, they often lack in quality.

Ask yourself: as a reader, would you rather read a million-word novel generated by a chatbot, or a beautifully crafted half-million-word novel written by a renowned author?

The choice feels obvious.

Time is not the equaliser here. Even if you trained a model on Shakespeare, the writing would still lack something essential. And that leads us to the next point.

AIs Are Rational, Not Emotional

Rationality is useful — it allows AI to infer patterns, synthesise facts, and respond with consistency. But this is also a trap. Inference is only as reliable as the information it is built upon. Feed an AI misinformation and it will confidently generate conclusions that wander far off the mark.

More importantly, people rarely read just for facts. They read for experience. Emotion is a central part of human expression, and while AI can mimic tone and style, it doesn’t feel anything.

This emotional gap is sometimes a strength, but often a limitation.

AIs Are Getting More Powerful Every Day

It’s true — AI today can do things that seemed impossible even a few years ago. But ask yourself: will there ever be a day when we trust AI completely?

Professional developers can’t blindly trust one another, let alone an AI system that occasionally makes baffling mistakes in areas you understand perfectly well.

If you can’t trust a tool in domains you do fully grasp, how can you trust it in domains you don’t?

So Should We Avoid Using AI?

Of course not. The key is to remain cautious, understand what AI is good at, and recognise its limits. Don’t use it to do things you don’t understand. Use it the way you’d use any tool: for empowerment.

AI as a Tool for Empowerment

A typical product development team includes product managers, designers, developers, and QA engineers. Each has distinct responsibilities and understands the work of the others at least to some degree.

• Product managers turn user needs into actionable tasks.
• Designers craft what the user ultimately experiences.
• Developers implement those designs.
• QA engineers seal everything with thorough testing and edge-case analysis.

Across this workflow, AI can take on the chores that slow people down.

If you’re a product manager, sorting a mountain of user feedback may feel like a chore.

If you’re a designer, understanding system constraints without constantly asking developers becomes far easier with AI acting as a technical translator.

If you’re a developer, writing clear PR summaries after a long day of coding may drain more mental energy than you’d like.

If you’re in QA, generating comprehensive edge cases can be time-consuming.

Each of these is an opportunity for AI to save time and reduce cognitive load.

So… Should AI Replace Our Colleagues?

In short: no. I don’t think it’s possible.

Ultimately, humans with domain expertise still make the decisions. Only developers truly understand the nuances of binding logic to UI. Only security engineers instinctively know never to commit API keys to a public repo. Only experienced practitioners recognise the countless tiny pitfalls that appear throughout real-world work.

AI can assist, but it cannot replace this expertise.

Conclusion

Returning to what my supervisor said, I would rephrase it like this:

‘I don’t know if AI will replace humans, but it will replace those who let AI think for them instead of using AI to help them think.’

Using AI to eliminate drudgery and accelerate ideas makes you more capable, not less.

But relying on AI to think in your place — to write code you cannot understand or to make decisions you cannot evaluate — leads to stagnation. You stop learning. You lose your judgement.

That is what makes a person obsolete — not the advancement of AI itself.