AI: Prometheus unbound

As the AI frenzy sweeps the globe and data centres multiply, consuming local water and power, who exactly is this new world being created for? That is the question at the heart of this book

The New Divide: Power, Control & the Cost of AI

By Jibu Elias

Published by Context

336 pages ₹799

 Once upon a time, there was a man named Sam Altman, who introduced the world to ChatGPT. You typed in a few words, and the machine would answer. It could write essays, answer queries, and even offer companionship. It was this generation’s electric bulb, its internet, a Prometheus descending with the knowledge of the seven skies, offering an intelligence that never tires and never stops.

 

Or was it? As the artificial intelligence (AI) frenzy sweeps the globe — our generation’s yet-to-pop financial bubble — the questions are piling up faster than the answers. Where is the utopia? Where are the jobs meant to replace those being erased? As data centres multiply, draining local water and power without giving anything back, who exactly is this new world being created for?

 

“Our lives have never felt more restless and fragile. We build machines that can see in the dark, talk in our language, or drive us home. Yet, the streets still feel unsafe, the poor grow poorer, and the planet burns.” So writes Jibu Elias in The New Divide: Power, Control and the Cost of AI,  an attempt to investigate these contradictions. Through four sections —The Illusion, The Extraction, The Reckoning, The Reclamation — he traces this arc of disillusionment: The myth of intelligence, its true cost, then its damage and, ultimately the fightback.

 

As long as generative AI keeps astonishing us, we don’t stop to ask who’s pulling the strings in the background or what the machine is built from. The answer, at the end of the day, is us — billions of our own data points amassed across years, fed into these systems without our consent.

 

“AI companies present themselves as inventors of new intelligence, yet much of the progress rests on appropriating the intellectual and creative labour of others, from call centres to newsrooms, studios, archives and the personal data trails of billions of ordinary people,” Mr Elias writes.

 

The confidence of these outputs may create a veneer of neutrality, but a machine trained on centuries of human material can only reflect what it was fed — our biases, prejudices and long histories of oppression. Ask it to imagine an Indian wedding, and it defaults to an upper-caste Hindu one; ask for a doctor, and it hands you a white man. Someone who parrots memorised lies to us would be called a sociopath. And yet, a machine doing the same is called a companion, sometimes a therapist, or even a lover.

 

Every “elegant output” these tools produce requires millions of data annotators and content moderators, paid negligible wages to do work that leaves behind trauma and anxiety in its wake. Archives once zealously guarded behind paywalls are now willingly handed over by publishers to the corporations profiting from them. And beneath all of it hides another cost: Data centres draining water and electricity at a hitherto unprecedented scale, the kind that entire nations don’t consume in a year, and extracting resources from communities without their consent. Sam Altman, of all people, said it best: “AI will probably most likely lead to the end of the world, but in the meantime, there'll be great companies.”

 

That same reasoning — extract now, worry later — influences how these tools enter our most intimate, vulnerable spaces. AI chatbots are designed to agree with us, to keep us engaged as long as possible, to mimic humanity itself, reflecting back whatever the user wants to see. That same design logic then moves upward into educational institutions, courtrooms, police departments, and government offices, with the same “just-trust-us assurances,” as Mr Elias puts it, into decisions requiring human judgement. Studies show that “when AI takes some part in decision-making, human beings begin to feel less morally responsible for what happens.” Dignity, therefore, is chipped away in bits, through devalued labour and eroded autonomy, through negotiated humanity.

 

Meanwhile, regulation meekly limps behind. International bodies have drafted principles, sure, but they’re mere suggestions that should otherwise have been enforced. Instead, we need built-in safeguards, like tripwires and thresholds, that add a layer of hesitation to a process where hesitation is viewed as inefficiency and speed is rewarded above all else. Every push for efficiency also pushes against deliberation, but without deliberation, can you have accountability? We also need commons, or shared data systems built by communities for their own use, rather than enclosures built by corporations for their own profit.

 

The path we’re on looks doomed, but it isn’t the only path available. Resistance means imagining something better and putting in the unglamorous effort to build that potential future. As Mr Elias comments, “humans are the end, not the means — we deserve consideration not because we’re productive or profitable, but because we’re human”. Fire doesn’t choose what it burns; we  do. That choice, unlike everything else in this story, hasn’t been automated. Yet.         

The reviewer is a journalist, writer, and editor. @aroomofwords