The cost of convenience: Ethics and AI in the Cook Islands

Photo: An average server room generates heat. So, imagine the heat put out by many acres of servers essential for AI content generation. They consume millions of litres of water per day, turning potable (drinkable) sources into non-potable chemical wastewater.

In island communities where access to graphic designers, writers, or artists can be limited—and time can be scarce—artificial intelligence has quickly become a quiet saviour. With a simple prompt, small businesses, sports clubs, community groups, and cultural organisations can instantly generate event posters, refine grant proposals, or compose background music. It is fast, accessible, and seemingly free. But beneath this seamless convenience lies an unseen footprint that warrants a closer look.

While we view AI as just another harmless digital tool, it has an unavoidable physical component. AI, especially when used for image and music creation, is astonishingly resource-heavy. The physical infrastructure behind it—sprawling data centres housing thousands of heat-generating servers—demands land space, generates noise 24/7, requires massive amounts of electricity, and uses millions of litres of water per day for cooling. This inevitably puts a strain on nearby communities through noise pollution, heavy electricity demands, and severe water degradation. In context, while our households receive 38,000 litres of clean drinking water per month for free, each data centre consumes millions of litres per day, turning potable (drinkable) sources into non-potable chemical wastewater.

Yet the environmental toll is only half the story; the human and cultural costs are equally profound. To train algorithms to filter out toxic content, violent imagery, and hate speech, tech companies routinely outsource data annotation to low-income countries. Thousands of workers in nations like Kenya and the Philippines are paid minimal wages to review graphic and psychologically distressing material for hours on end, meaning these people often suffer long-term trauma—all so that users elsewhere, like us, will enjoy a ‘safe’ digital experience.

At the same time, generative models (creates new content like images or music—rather than finding existing information) are indiscriminate, consuming the uncredited work of human creators while making our visual world all look the same. On social media, we regularly see AI-generated posters featuring designs that, while vaguely Polynesian, reflect Tahitian or Samoan motifs, alongside generic island outlines that look more like Hawaii than Rarotonga or any of our pa enua. When we rely on AI for imagery, we do not just skip supporting a local artist; we are actively eroding our unique identity and starving our own creators of the opportunity to earn a living, express their heritage, and pass down their skills.

This raises a crucial question for remote communities: when a digital tool’s supply chain causes environmental harm, human exploitation, and cultural erosion, what is our social responsibility?

Much like the internet itself, AI is neither inherently good nor bad; its impact depends entirely on how we choose to use it. Most of us use the web in work, or to connect and learn, while striving to avoid its darker corners. AI requires that same thoughtfulness or intent. For isolated island organisations, total avoidance is likely impractical, but developing a clear ‘Responsible AI Use Policy’ is becoming essential.

For those who care, perhaps it is time for local teams to establish clear internal boundaries/policies for ‘ethical digital stewardship.’ Some simple, practical choices could be:

Ultimately, convenience should not cost us our conscience or our culture. Genuine stewardship means ensuring our values hold true everywhere—living aligned with who we are at home, while also respecting the environment and unseen human lives we touch, intentionally or not, across the globe.

(Gemini AI and LM Studio were used in creating this article)