Go ahead, put data centres in space—but AI had better improve lives back here on Earth

This week, Silicon Valley venture capitalist Vinod Khosla made an even more bold prediction. Speaking to Prannoy Roy on DeKoder, he laid out a vision so utopian that it resembled India as a paradise. Khosla prophesized that by 2030, a poor child in rural India would, via AI, have tutors as good as those of wealthy children in big cities. A farmer, Khosla said, will have access to doctors via AI who will somehow know his entire medical history.

This space odyssey of starry-eyed scenarios confusingly played out against reports of chaos on the opening day at the India AI Impact Summit in New Delhi.

The long queues caused by contradictory and complicated instructions from security personnel were recognizable to anyone who uses our otherwise glamorous airports. Even so, the shortage of drinking-water refill stations on site reported by many attendees and the instructions not to bring car keys to the venue must rank as an Olympian embarrassment.

A ‘tip sheet’ for attending the AI Summit doing the rounds wittily suggested, “Hydrate like you’re crossing a desert. Drink 2 litres of water before leaving so you don’t have to hunt for water there. After parking, hide your car key under a random brick or stone. Bring enough cash because technology might take a day off.”

I am a Luddite who prefers vinyl and compact discs to streaming and would be part of the QuitGPT movement but for the fact that I have never used ChatGPT. Yet, having lived in Delhi almost a decade and a half ago, the code I learnt early on was one our AI startups overlooked: Be wary of Delhi jamborees and summits. It is better to stay in office and work a 12-hour day than navigate the capital’s VIP movements to attend such an event.

Indeed, what the AI Summit demonstrated is that the ground realities of dealing with the government, the blessings of UPI and online tax-filing aside, have not changed as much as they should have even as we have supposedly leapfrogged to the 21st century. The large language model that the Indian government needs is a digital interface with its 1.4 billion citizens where the technology and task execution proceed seamlessly.

A friend in Bengaluru this week recounted his 85-year-old mother’s travails trying to claim her pension via the Jeevan Pramaan life certificate scheme for seniors. The biometrics are a hindrance, no matter how often she blinks at the camera as instructed by the software. An Aadhaar interface has left her stressed and worse off; her pension is now three months in arrears.

This week, an aunt in Bengaluru sought to register a trust and a will, both intended to help manage finances for her schizophrenic sister. Weeks in advance, the three signatories committed to being in town on Wednesday. The sub-registrar’s office server was malfunctioning, however, with no OTPs being generated. Aadhaar details for KYC could not be uploaded. Armed with mandatory self-attested paper copies of Aadhaar and PAN cards and the obligatory passport photos of yesteryear, the signatories had no way to present them.

Even as both the state and private sector struggle to give up their love for the eternal paper chase, the true test for AI is tools such as Sarlaben, which helps dairy farmers in Gujarat, and MahaVistaar, for farmers in Maharashtra. As Mint’s uplifting Long Story on Tuesday recounted, both are successes. The story quoted a sugarcane farmer who used an AI app to drastically reduce the chemical he was using as wrongly advised by a village shop. The red spots on his sugarcane disappeared. When a cow is sick, there is no need to wait for a vet.

“My Sarlaben. Amul AI” is far more responsive. As these apps achieve scale—and the speed with which they are being created and used is impressive—the increase we will see in productivity will be far more dramatic than the occasional days lost waiting for government servers behaving whimsically.

Yet, buried amid the headlines this week about how AI will change the world was an alarming story about how bright orange water is flowing in rivers north of the Arctic Circle. This is a reminder that for all that AI can do in the present and future, we are not making much progress in arresting climate change.The culprit, Financial Times reports, is global warming: Thawing permafrost is resulting in “metals leaching into water systems.” Another study linked air pollution to a higher incidence of Alzheimer’s.

Still, there is some good news even on climate change. Overlooked in all the hype about data centres is how much energy and water they use to power and cool them. Placing gigantic data centres in space, which has the virtue of being very cold, will at least reduce their otherwise heavy contribution to climate change. At least that is what Perplexity assures me.

The author is a former Financial Times foreign correspondent.