New Delhi
AI is not quite available to all, according to Prof Pradeep Khosla, Chancellor of University of California San Diego. The version of ChatGPT available in the US is 10x better than that available in India, he points out. “Every country needs to think about its AI strategy,” he says.
Dr Khosla was speaking at AIMA’s National Leadership Conclave.
He said that not all countries can afford to train their own large language models because of the high costs and the governments will have to invest in AI like infrastructure. He said that countries will need to collaborate and cooperate in using AI, just as they do in the case of defence. He said that not all countries would have the best AI and pointed to their vulnerability in an era of AI-powered war drones.
Talking about general application of AI, Prof Khosla pointed to healthcare and education as key use areas. He said that while technology has greatly impacted the per unit cost in most industries, the cost of delivering education has not gone down despite technological advances. He pointed out that the cost of tuition goes up faster than inflation and the majority of American families cannot send their children to college because of high cost of university education. AI will made a dent in the cost of education, he said referring to UCSD’s experimental AI tutor which is 99% more efficient in cost per unit of education delivered compared to a human tutor.
In the healthcare sector, Prof Khosla said that AI will drive a shift from episodic monitoring of health to continuous monitoring and from post-illness consultation to predictive advice. He mentioned that UCSD’s algorithms have reduced the hospital deaths caused by sepsis by 30%.
On the fears of AI creating its own social and economic problems, Prof Khosla said that AI is only as good as the data it is trained on. There can be criminal and lying AI models too, just like people, and the world would need a new legal regime for AI and software. “For the parallel universe of AI agents, we have to create different laws,” he said.
On the concern that reliance on AI will create a problem of uniformity and mediocrity in education and work, Prof Khosla said that AI is merely a tool and the output will depend on the individual skills with the tool. Also, there will be different kinds of AI tools for different kinds of work.
“A combination of humans and AI will be more perfect that either of them can be on their own,” he said.
The session was moderated by Mr Nikhil Sawhney, President, AIMA.
Mr Sawhney expressed concern about variation in AI regulation in different countries. He pointed out that Chinese large language models have been permitted to break down global firewalls and obtain data.
Prof Khosla said that new technological breakthroughs always raise ethical concerns but the world gets used to the change. He pointed out that when a goat was first cloned, the technology was opposed in the US but the Chinese went ahead with it.
On controlling, or stopping, the AI, Prof Khosla said that computers need electricity to work and one can always pull the plug.
The session was live streamed on AIMA’s social media channels.
Quick 
Submit Enquiry