WHILE scrolling through social media, I recently saw a post that made me pause. The questions were actually quite simple: Why are artificial intelligence (AI) tools and robots not used for jobs that are dangerous for humans? Why do we use AI and robots to write our emails, prepare our presentations and do the thinking that humans can do well on their own? Why not send robots into mines? Why not use robots to handle chemicals, work in extreme temperatures, enter disaster zones, or take up tasks that can seriously harm human health?

The International Labour Organisation (ILO) has made the argument itself, stressing that robotics, automation and AI can work in tricky environments and take on dangerous tasks, such as handling toxic materials or working in extreme conditions. Mining, for instance, remains one of the world’s hazardous occupations. In Pakistan, ILO has often highlighted the safety challenges that are faced by mine workers, with hundreds of fatalities having been reported nationally in recent years.

So, then, why is AI used for the opposite purposes? Why do we use AI to write our thoughts, formulate our opinions, prepare our presentations, compose our messages, and brainstorm for us before we have even had the chance to brainstorm ourselves? Let us think on our own. Let us struggle with an idea for a while. Let us write something badly before we learn how to write it well. Let us put our thoughts on paper, half-formed, messy, emotional, contradictory, sometimes even wrong.

There is a beauty in all that. Our writing carries our background, experiences, prejudices, humour, memories, education and our worldview. Even when the writing is imperfect, it tells us something about the person behind it.

AI tools can surely make a paragraph polished, but in doing so they sometimes make everything sound strangely alike. This is not just a philosophical concern. Research is now beginning to show it. A 2025 study published in the Nature Human Behaviour found that while ChatGPT could improve the creativity of ideas, relying on GPT reduced the diversity of ideas within a group, which is a critical quality in brainstorming sessions. Another 2025 study that compared thousands of human and GPT-generated essays found that human-written essays contributed new ideas with the diversity gap widening as more AI-generated writing was introduced.

There is also a cost worth considering. A recent study done at the Massachusetts Institute of Technology (MIT) examined AI-assisted essay writing and raised valid concerns about what happens when people repeatedly outsource parts of the thinking and writing processes to AI. Perhaps the question we should ask is not ‘What can AI do for us?’ We already know the answer to that. The question should be reframed as: ‘What should we let AI do for us?’ Let technology minimise risks for humans. Let machines handle repetitive and physically damaging work.

Let AI process datasets, detect patterns, automate routine administration, and help us solve problems beyond our natural limitations. When it comes to thinking, imagining, questioning, creating and making meaning, we should keep that task for ourselves. Technology should not be allowed to make humans unnecessary; it should give humans room to be human.

Maryam Zahra
Karachi

Published in Dawn, October 20th, 2026