More statistics PhDs needed for India to advance in artificial intelligence
Prasanta Chandra Mahalanobis; courtesy WikiMedia Commons
By Ashok Kumar Nag*
March 21, 2022
In 1976, after completing a five-year combined Bachelor’s and Master’s degree program at the Indian Statistical Institute (ISI), Kolkata, I was accepted into the PhD program also at ISI. At that time, ISI had two centers – Kolkata and Delhi, where Ph. D students were admitted. There were a few other centers for running training and consulting programs on Statistical Quality Control for industries.
Today, ISI has 5 campuses at Kolkata, Delhi, Bangalore, Chennai and Tezpore (Assam) as well as a branch office at Giridih, in Jharkhand state, for conducting statistical research related to agriculture. In addition, ISI has Statistical Quality Control and Operations Research units in Coimbatore, Hyderabad, Mumbai, and Pune.
Today, forty years after I earned my Ph.D. at ISI, I find that, despite its geographical spread, the institution has lost its focus and global stature. This is evident from various rankings of similar academic institutions.
For instance, in the 2021 Shanghai Ranking’s Global Ranking of statistical research institutes, ISI has received a score which is one tenth of that of Harvard University, which was ranked number one. The data compiled by the scientific publication database Scopus, covering the years 2011-2020, shows that the per faculty publication by the Statistics and Mathematics department at the ISIs was the lowest among their peer groups.
The major reason for the decline of ISI is the shift in its focus to computer science, away from statistics. During the past five academic years, 2017 to 2021, the institution awarded a total of 114 PhD’s. Only 18 of those were in statistics. Another 26 doctorates were awarded in mathematics. During the past five years, the major share, 48, of the PhDs were awarded in computer science. But this shift has been underway for over a decade. In 2010-11, for instance, only one out of 11 PhDs were awarded in statistics, as compared to three in computer science.
Yes, India needs more talent with advanced computer science education. But that ought to be achieved by raising the number of PhDs from the 23 Indian Institutes of Technology. The Computer Science department of IIT Bombay alone has an annual intake 20 Ph. D students. Also, the IITs have the faculty and facilities for advanced computer science research, since they collectively enroll about 1030 students each year for their four-year undergraduate degree programs in computer science. It is, therefore, more productive to add to the faculty and facilities at the IITs than shift the focus of a specialized institution like the ISI.
The rise in the number of computer science PhDs at ISI is also reflected in the far higher number of faculty hired in the computer science related departments, while there is a decline in the number of faculty in the statistics departments. But this was not the vision of ISI’s founder.
The origins of ISI can be traced to 1931 when Professor Prasanta Chandra Mahalanobis founded the statistical laboratory at Presidency College, Kolkata, now Presidency University. He was a product of the Bengal Renaissance, a reformist movement in early 20th Century British India. Its goal was to liberate Indians from the shackles of mythic beliefs and ancient rituals in order to advance society through the pursuit of rational knowledge and methods.
Mahalanobis sought to apply statistics to a wide area of study. At ISI, he set up statistical research units in biostatistics, anthropometry, botany, agriculture, human genetics, embryology and biochemistry. He invited world class mathematicians and statisticians, including R.A. Fisher, Norbert Wiener and W. Edwards Deming, as well as noted British biologist and geneticist J.B.S. Haldane, to conduct research and teach ISI students.
As a result, ISI attracted the best students and faculty in statistics in India. It was also an influential institution with projects ranging from organizing the first nation-wide sample surveys of income, poverty and other economic data, in 1950, to formulating India’s second 5-year economic plan, 1955-1960.
Today, the decline of the ISIs comes at an inopportune moment for India. There continues to be strong demand for good quality statisticians and statistical research methods in traditional areas like banking, finance and policy formulations, as well as in new fields like data collection and analysis at internet businesses.
More important, the development and use of artificial intelligence (AI) tools are expanding rapidly, from commercial applications from retail to robotics to analyzing satellite data for defense purposes. Both in terms of AI tools and talent, India is far behind the U.S., U.K, Canada and, most worrisome, China.
In 2019, 180 colleges in China were teaching AI and related majors for undergraduate students. More than 60 universities in China offer Masters’ and Ph.D. programs in AI. Each year they graduate over 1,200 students with advanced degrees in the field.
Globally, most of the key AI talent comes from those with advanced skills in statistics and mathematics. “Statistics, as an interdisciplinary scientific field, plays a substantial role both for the theoretical and practical understanding of AI and for its future development,” according to a paper published by a team of German researchers last year.
A vision similar to Mahalanobis’ is required at the ISIs today. Instead of continuing to shift to computer science, they need to enroll more students and faculty for their statistics and math programs.
Currently, there are at least 15 researchers at ISI working in the field of AI and related Image Processing domain. Using their skills, the institution has the potential to boost the development of AI in India, as it did in the field of statistics.
The various ISI campuses should collaborate to share faculty and facilities as well as invite the top global AI experts to teach and initiate research. While experts based abroad may find it tough to travel to India, due to their tight schedules, they should be invited to conduct lectures and seminars online. In fact, lectures, talks, seminars and other public discussion by leading experts should be simultaneously broadcast online to all ISI campuses. This will get more students interested in the study of advanced statistics, mathematics and AI.
There is a global shortage of AI talent. Getting admitted to a PhD program in AI at Cambridge University, the University of Toronto or MIT is far tougher than getting into a computer science Ph.D. program at the same institutions. This is because PhDs from the top AI institutions in the western countries – as well as in China - are commanding salaries and bonuses far higher than that paid to computer science PhDs.
The giant American companies, for instance, seek to lure the best AI talent from around the world. They offer a starting annual salary of $500,000, plus a large signing bonus, for a top quality Ph.D. in AI from Cambridge or MIT. As the number and quality of the PhDs earned in statistics and math rise at the ISIs, it is likely that Google, Microsoft and other leading Western technology companies will fund the expansion of faculty, research and students at the ISIs – to meet their own need for talent.
In 1950, in his presidential address to the Indian Science Congress, Mahalanobis noted that there cannot be any separation “between the theory and the practice of statistics.” If the ISIs turn their focus back on statistics, and contribute to the evolving practical applications of AI, they will likely play a bigger role in India’s technological future than the IIT’s.
For weekly stories follow on: LINKEDIN or TWITTER or FACEBOOK
Subscribe and support Global Indian Times via THIS LINK
*Ashok Kumar Nag is a Mumbai based consultant in information management and data analysis. He spent over 25 years in the Statistics and Information Management department of the Reserve Bank of India, retiring as an adviser. Ashok is an alumnus of the Indian Statistical Institute, Kolkata, India. Link
