Role of AI in health discussed
The future of artificial intelligence was not just discussed on the 2nd day of the AI-Impact Summit — it was put to an immediate vote. In a lively session led by Professor David Yanagizawa-Drott, delegates were asked to choose how a hypothetical Indian State’s health system should deploy AI in district hospitals: rely solely on doctors, hand decisions entirely to machines, or combine the two. At first, nearly 89 per cent backed “augmentation”, with AI assisting doctors who retained final authority.
Even when told automation could cut labour costs by 25 per cent compared with 20 per cent for augmentation and none for the status quo, most stuck with the hybrid model. But when new evidence suggested AI-only systems might worsen patient outcomes by 30 per cent, while augmentation risked a 5 per cent decline, support shifted dramatically.
Backing for human-only care rose to about 40 per cent, augmentation fell to 56 per cent, and full automation collapsed to 4 per cent, a vivid illustration of how quickly opinion turns when the evidence changes.Drawing on field research in Ghana, Professor Yanagizawa-Drott described a real-world trial in which an organisation recruiting teachers for rural schools tested whether to use GPT-4 to automate candidate screening or combine its recommendations with human judgement.
Contrary to popular belief, augmentation proved the weakest option, slowing the process without improving results. Full automation, in that instance, both reduced costs and improved the quality of hires. The findings challenge the comfortable assumption that human–AI collaboration automatically delivers the best of both worlds. “Researchers cannot give you the values,” the professor told delegates, “but we can help you establish the facts.” In a fast-moving technological landscape, he argued, policy must be guided by evidence rather than instinct.
The subsequent panel broadened the debate beyond hospitals to jobs and livelihoods. Citing recent World Bank research, speakers pointed to productivity gains in India’s IT sector alongside continued job creation, pushing back against predictions of mass technological unemployment. Shankar Maruwada of A Step Foundation highlighted a ‘Blue Dot’ model designed to connect informal workers with local opportunities through digital tools.
Yet panellists also warned that AI’s benefits may not be evenly shared, noting emerging gender and age gaps in the use of generative systems. As discussions at Bharat Mandapam made clear, the question is no longer whether AI will shape the future of work, but how Governments can ensure its rewards are widely distributed and firmly grounded in proof rather than presumption.
