In the past year, there has been a widespread, wobbly feeling of the ground shifting beneath our feet. It’s been a period of exponential change and extreme uncertainty, all sparked by the launch of a single digital product: ChatGPT, a website that can respond with detailed answers to conversational queries. The chatbot had a simple interface — a blinking cursor that said ‘send a message’. It was our first direct interaction with artificial intelligence software. Within days of its release in November 2022, ChatGPT captured the imagination of millions of people who were enthralled by its ability to ‘talk’ back to us using nuanced written language. It has since been co-opted by 100 million users a week, from businesses and law firms to software engineers and students.
This has marked a profound shift in our relationship with machines, involving mind-bending questions such as: who owns the rights to all of humanity’s creative outputs and can they be recreated and re-sold by AI software? Will white-collar jobs continue to exist as they are? And will so-called knowledge workers such as lawyers, journalists, consultants and creative professionals still have work in a few years’ time, when their jobs could be completed by generative AI to a good enough standard?
There’s the knock-on question of how a society can be sustained without work, and concerns about the future of our species: how will children learn while leaning heavily on these tools? Can they think properly without learning to write well? And, ultimately, who the hell are we if all our ideas and thoughts can be replicated by machines?
As scientists, technologists, philosophers, economists and politicians try to answer these questions for the public, all types of AI — from generative software to decision-making systems — are catching on like wildfire, racing on through the economy at a pace much faster than any government trying to contain it.
Leaning on algorithm
Author Madhumita Murgia
Keeping up with developments in AI requires observing how it is diffusing into every aspect of our society. We lean on algorithmic technology just as we once did on each other. When you open Google Maps, call out to Alexa, book an Uber, or listen to a recommended song on Spotify, you are dealing with a form of AI. The content on your social feeds and the ads you are served for beach holidays or children’s clothing are targeted at you using AI. When you try to get a loan from a bank, you are screened by AI. What price you pay for your home, or your car insurance, are decided by AI. When you are interviewing for a job, your face and responses may be analysed by AI. Maybe you even used AI to write your job application.
The outputs of AI software today can help human experts make consequential decisions in areas such as medical diagnosis, public welfare, hiring and firing, among others. The area of healthcare is one where AI can have truly life-saving potential — which is why Ashita Singh has been using it for the past few years.
In 2019, when Dr. Singh first heard about an AI program designed by a Mumbai-based company called Qure.ai, that could help to diagnose tuberculosis, she was sceptical. Dr. Singh was one of eight doctors at the Chinchpada Christian Hospital — a tiny mission hospital run by the Emmanuel Trust in the district of Nandurbar, a cluster of tribal villages on the border of Maharashtra and Gujarat. The majority of Dr. Singh’s patients are Adivasis, belonging to the Bhil tribe. The Bhils needed the basics: better nutrition and housing, access to more qualified doctors and early diagnoses, and common life-saving drugs like antibiotics. Dr. Singh felt that technology, such as digital medical devices, often failed her patients because the systems were not designed for remote and under-resourced settings like Chinchpada.
Helping rural folk
Dr. Ashita Singh (right) visits a patient at home in Nandurbar, Maharashtra. | Photo Credit: coalitionagainsttyphoid.org
But then, her curiosity was piqued, and she began reading about Qure’s AI system, known as qXR. She learned that the software was trained to pick up visual patterns suggesting tuberculosis in X-rays and output a patient’s potential risk for it. Other trials had shown the app performed on par with the best radiologists. A suspected tuberculosis patient would still need medical confirmation, and a human was needed to deliver a diagnosis and prescribe treatment. But if they could read X-rays accurately, she believed AI systems could be transformational for rural folk.
“We have people travelling to us five or six hours each way for a straightforward diagnosis of TB, simply because of the many months they spent going to quack after quack, getting IV fluids and cough syrups,” she tells me. In the previous week alone, she’d had five patients admitted on a single day with advanced TB. Two had died. This is the key to implementing AI technologies successfully — to find where they can truly make a difference. “We aren’t talking about improving quality of life here, we are talking about preventing death,” Dr. Singh says. “Imagine what a gift it would be.”
The writer is Artificial Intelligence Editor at the Financial Times, and author of the forthcoming book Code Dependent: Living in the Shadow of AI (Picador), longlisted for the 2024 Women’s Prize for Non-Fiction.
