Who Owes Teresios Bundi Anything?

Teresios Bundi's story is not really about whether AI is good or bad - it's about who pays when a technology makes an entire category of human labour worthless almost overnight and whether anyone owes the people who pay that price anything at all.

For twelve years, Bundi built a livelihood and eventually a small business employing other Kenyans, writing academic essays for overseas students. None of that labour was invisible or accidental - it was skilled, sustained work that supported thousands of families across Nairobi.

When ChatGPT arrived in 2022, the value Bundi and 40,000 others had been selling - competent, on-demand written text - became available to anyone, instantly, at near-zero cost. Within about two years, the industry that had employed tens of thousands of people had shrunk to almost nothing.

Here is the concrete ethical tension: the companies that built and released the AI models captured enormous value from replacing exactly this kind of labour - efficiency gains, market dominance, profit - while bearing none of the transition costs their disruption caused.

Bundi and his peers received no notice, no severance, no retraining support and no seat at any table where this transition was being planned, because there was no formal employment relationship connecting them to OpenAI at all.

The company owed him nothing under any contract, because no contract ever existed between them. That absence of a legal relationship is precisely what makes the ethical question sharper, not softer - a company can cause devastating harm to people it has never employed, never negotiated with and never even known existed, simply by releasing a product into a global market.

The utilitarian case for generative AI is genuinely strong: it has created enormous aggregate value for hundreds of millions of users and unlocked capabilities with real social benefit. But distributive justice asks a different question than "does this create more total value" - it asks who specifically bears the cost of getting there and whether that allocation is fair.

A calculation that produces net positive utility globally while concentrating catastrophic, uncompensated loss on a specific, identifiable, already-vulnerable population is not automatically justified just because the aggregate math works out.

The honest resolution here does not pretend this dilemma disappears. Ethical responsibility for large-scale technological disruption should be treated as real and owed, even across a border no contract ever crossed, precisely because impact - not a signed agreement - is what generates moral obligation.

This doesn't mean AI companies bear sole responsibility or that governments and international bodies have no role; but a framework that lets the party capturing the gains disclaim all responsibility for the losses, simply because no contract linked them to the people bearing those losses, is not a neutral default - it's a choice that happens to favour whoever already has the power to make it.

Something is genuinely sacrificed either way: holding companies to this standard could slow beneficial innovation, while refusing to hold them to it leaves workers like Bundi with nothing but the advice he now gives others - learn to survive the next disruption alone.