Synthetic intelligence (AI) is more and more being framed as a path to financial development, strategic autonomy, and nationwide energy. However what occurs when the prices of that ambition are borne by employees, communities and public assets that stay largely invisible?
The “AI Resist Listing” seeks to reply that query. The venture paperwork international efforts to problem and reshape the growing influence of AI throughout areas starting from labour and knowledge extraction to surveillance and digital infrastructure.
Petra Molnar, a lawyer and anthropologist who research AI, surveillance, and human rights, and is a part of the venture, instructed The Indian Specific why narratives of AI-driven development typically obscure questions of labour, surveillance, infrastructure, and democratic accountability.
Many nations, together with India, are embracing AI as a pathway to development, competitiveness and nationwide energy. How did AI adoption turn out to be such a extensively shared political objective?
The story of AI as a nationwide crucial is, at its core, a narrative in regards to the seductiveness of technological shortcuts. For governments navigating post-colonial legacies, widening inequality, and the anxieties of geopolitical irrelevance, AI provides a solution to skip the messy, sluggish work of structural transformation and arrive at a future that appears like development via algorithmic effectivity. The US and China modelled this playbook aggressively, and the strain to comply with has been huge within the so-called AI Arms Race, with nations jostling for AI supremacy.
What the AI Resist Listing reveals is that this “AI as improvement” framing is itself a type of narrative energy. It is without doubt one of the 4 pillars we doc: Resist, Refuse, Reclaim, and Reimagine. The narrative infrastructure of AI — the tales instructed about what AI is for, who it serves, and what progress appears to be like like — is as materials as the info centres and chip provide chains. Nations like India have absorbed not simply the know-how however the concept that energy within the twenty first century requires AI dominance, and that resistance stands in the best way of progress.
This framing obscures what truly drives these selections: defence procurement pursuits, surveillance state consolidation, investor seize of public infrastructure, and the very concrete political utility of showing trendy and as an “AI Chief”.
Story continues beneath this advert
The checklist breaks AI energy into interconnected pillars: labour, infrastructure, surveillance, knowledge, and narrative. India seems throughout all of them. What does India’s place inform us?
India’s look throughout all of the pillars is not any coincidence. AI energy is assembled via extraction at a number of ranges, and no nation is just a sufferer or a beneficiary. India occupies a revealing place: it’s concurrently a website of exploitation and a website of aspiration.
On labour, India provides a lot of the invisible workforce that makes AI techniques legible. On infrastructure, knowledge centres are being pitched as improvement anchors whereas drawing on scarce water and electrical energy. On surveillance and knowledge, facial recognition techniques, networked CCTV, and the huge reserves generated via Digital India elevate important questions on oversight and consent.
Thus, AI energy shouldn’t be assembled in Silicon Valley after which shipped outward. It’s assembled via relationships which might be extractive and unequal. AI’s international provide chains rely on hierarchies which have structured capital for generations.
Story continues beneath this advert
AI firms typically current knowledge centres and laptop infrastructure as engines of improvement. What questions ought to communities and policymakers ask earlier than accepting these guarantees, particularly in nations dealing with water stress, electrical energy pressures and uneven improvement?
In my thoughts, the only most vital query is: who bears the prices, and who receives the advantages? For instance, knowledge centres eat huge portions of water for cooling and electrical energy that, in water-stressed and power-insecure areas, comes immediately on the expense of households, farmers, and small companies. The promised advantages of jobs, tax income and a stronger place within the digital financial system are inconsistently distributed. The engineers who workers these services usually are not from the communities that lose entry to water.
View of the Yotta D1 constructing’s rooftop chiller items on the complicated in Larger Noida on March 2, 2026. Photograph: Tashi Tobgyal
Policymakers ought to ask whether or not commitments are binding, whether or not communities can refuse or renegotiate tasks, whether or not environmental and social impacts have been independently assessed, and whether or not knowledge governance and know-how switch preparations genuinely profit native communities.
India desires strategic autonomy in AI whereas remaining depending on American chips, cloud suppliers and frontier fashions. Are genuinely sovereign AI ecosystems realistically attainable for nations exterior the US and China?
Story continues beneath this advert
Within the brief and medium time period, in all probability not in any complete sense. The infrastructure that underpins AI, from frontier fashions to semiconductor provide chains, is so deeply concentrated that formal sovereignty coexists with profound dependence. A rustic can localise knowledge and construct its personal massive language mannequin whereas nonetheless counting on overseas cloud infrastructure and Nvidia chips topic to US export controls.
This isn’t a cause to desert sovereignty as a objective, however it’s a cause to be exact about what sovereignty would truly imply. Probably the most achievable types of AI sovereignty are in regards to the capability to manage, audit, and refuse.
India has turn out to be central to AI knowledge labelling, moderation and annotation work. Why does the AI business erase the labour that makes these techniques attainable?
As a result of erasure is structural, not incidental. The magic trick of AI, and the rationale it instructions the valuations it does and the political creativeness it captures, will depend on the looks of automation. In case you might see the employee in Hyderabad who spent 12 hours immediately labelling photos of automobile accidents for a self-driving system, or the moderator in Bengaluru who reviewed hundreds of items of graphic content material this week to maintain a platform clear, the automation thesis collapses. The labour needs to be made invisible for the product to operate as promised.
That is knowledge colonialism in its most concrete type: the extraction of cognitive labour from the World Majority to construct techniques whose earnings accumulate within the World North, beneath circumstances that echo colonial relationships via low wages, weak labour protections, and little management over how the techniques are used. The business’s silence about tech employees is a deliberate alternative, not an oversight.
Story continues beneath this advert
Do you see AI labour in nations like India as a brand new model of digital outsourcing, or one thing structurally totally different?
Structurally totally different, and I believe extra extractive in particular methods. Earlier outsourcing waves, from name centres to coding, concerned seen features that appeared in company accounting.
Nonetheless, AI knowledge labour is extra insidious as a result of it’s designed to vanish into the product. The employee’s contribution is absorbed right into a mannequin, flattened right into a coaching dataset, laundered right into a benchmark rating. There is no such thing as a second at which the system broadcasts {that a} functionality exists due to work carried out by a particular human being beneath particular circumstances.
There’s additionally a deeper query of energy asymmetry. Earlier outsourcing relationships, nevertheless unequal, no less than positioned the outsourcing firm as a visual counterparty with some accountability to contract phrases. A lot AI annotation work immediately is organised via platform-mediated gig constructions designed to keep away from employment relationships, leaving employees with fewer protections, much less collective energy, and little visibility into how their work is used.
Story continues beneath this advert
What types of accountability efforts point out that the present trajectory of AI improvement shouldn’t be inevitable?
The database (of the AI Resist Listing) paperwork efforts starting from authorized challenges in European courts to group refusals of facial recognition in public housing and employee organising throughout annotation provide chains.
What offers me hope is the range of those efforts and the diploma to which they’re related, even when the folks concerned have no idea one another. A group within the US refusing a facial recognition contract and an activist in India difficult the Digital Private Knowledge Safety Act are each insisting on the identical precept: that the deployment of know-how requires democratic consent and a real chance of refusal.
Whether or not AI turns into democratic infrastructure will rely on whether or not establishments can present significant accountability and whether or not communities retain the capability to problem applied sciences that have an effect on their rights.