2 April 2018

India takes the first step to building an AI vision

R. Shashank Reddy
Last year, the Union ministry of commerce constituted a task force to look at how Artificial Intelligence (AI) can be leveraged for India’s economic growth. The report of this task force, released recently, makes for interesting reading. The report focuses primarily on figuring out where AI technologies can be the most beneficial for the Indian state, economy, and citizens. To this end, it identifies 10 specific domains for rapid AI incorporation: manufacturing, fintech, health, agriculture, technology for the differently abled, national security, environment, public utility services, retail and education. Within these domains too, the report also identifies four “grand challenges” for AI incorporation: improving manufacturing, especially in the SME (small and medium-sized enterprises) sector; improving healthcare quality; improving agriculture yields; and improving delivery of public services.


In essence, the report is somewhere between a vision document and a semi-concrete policy proposal. A number of issues, such as AI education, have been dealt with in general terms, underlining the importance of this particular aspect but stopping short of spelling out in detail the manner of achieving these aims. On the other hand, as discussed below, the report does provide specific recommendations, each of which can be taken as a concrete policy proposal.

It is also necessary to point out three specific aspects of the report that really stand out. The first is the honest assessment that “the most important challenge in India is to collect, validate, standardize, correlate, archive and distribute AI-relevant data and make it accessible to organizations, people and systems without compromising privacy and ethics.” The availability of good quality data is indeed one of the biggest issues facing the development of AI in India, and remedying this must take top priority. The report proposes to do this via properly governed and regulated digital data marketplaces and exchanges.

The second is the importance given to the use of AI as an accessibility technology. While research in the use of AI to provide a better life for the differently abled is widespread, this is probably the first time that it has been highlighted as a core issue in a government document—an explicit policy recognition of the good AI can do.

Third has been the willingness of the task force to engage with the difficult AI policy questions. An entire section has been dedicated to the ethics of incorporating AI technologies, where the report bats for explainability in AI behaviour, rigorously audited data sets to prevent bias, and, in a very forward-thinking manner, calls for a discussion on the rights and responsibilities of autonomous entities. The report also repeatedly stresses the need to have better, and more effective, data protection policies as a precursor to encouraging any form of data sharing.

Another entire section tackles the vexing question of AI and employment. While, by its own admission, the task force is “sanguine” about the overall impact of AI on employment, the report argues, citing a number of a studies, that AI will in all likelihood create more jobs than it will destroy. However, the report also categorically states that India-specific quantitative models and studies need to be encouraged to arrive at informed conclusions about the relationship between AI and jobs in India.

On the concrete policy proposal side, the report provides a list of recommendations to the government of India, which include the establishment of an inter-ministerial National Artificial Intelligence Mission, creation of six centres of excellence for AI research, and the establishment of data banks and exchanges along with an ombudsman. Of these, arguably the most practical at the moment, and the one which might have the most long-lasting impact, is the creation of centres of excellence which can provide the foundation to encourage greater AI research and development in India.

There are, however, a couple of issues with the report. The first is the perfunctory manner in which the issue of AI and national security has been dealt with. Presumably, this could be because there also exists a ministry of defence task force (of which I am a member) that has been constituted to look at this specific issue in detail. However, given the fact that the use of AI in national security will necessarily depend on close relations between the military and the private sector, it would have been beneficial to read the task force’s views on how this can be brought about.

Second, while this report is certainly very welcome, it remains to be seen how the work of this particular task force fits into the larger role given to the NITI Aayog to develop a national AI strategy. Prima facie, the focus areas of this task force and NITI Aayog’s AI mission seem to be similar. To prevent a duplication of efforts, therefore, and given the thought that has gone into the preparation of this report, NITI Aayog should build the recommendations given by the task force into its deliberations.

Overall, this report is a very welcome step that highlights the government’s readiness to tackle the policy issues and opportunities thrown up by emerging technologies. While the effectiveness of this report can only be gauged by the manner in which its recommendations are adopted, hopefully it will kick-start a much wider discussion on AI policy and attendant issues in India.

R. Shashank Reddy is a research analyst at Carnegie India.

No comments: