NASSCOM feedback on the Discussion Paper on Indian AI stack
Blog: NASSCOM Official Blog
On 2nd September 2020, the AI Standardisation Committee published the Discussion Paper on the Framework of an Indian AI stack for public consultation. The paper proposes a stack that seeks to remove the impediments to AI deployment by putting in place a comprehensive framework.
In this context, on 15th October, NASSCOM had organised an interactive meeting between the Committee and AI experts from the industry. Chairman, AI standardisation Committee made a presentation on the proposed framework and shared the approach the Committee wanted to follow. The presentation was followed by an open discussion, where the members from the industry shared their view on the proposals.
Further, NASSCOM made a submission with detailed feedback on the proposal. The key points made in our submission are listed below.
- Currently, a multitude of government agencies/departments – MEITY, MOC, Niti Aayog, DOT – are working on various aspects of AI regulation. However, we believe the efforts of the Govt of India can benefit by becoming cohesive. This cohesion may be achieved by developing a coordination mechanism between different arms of the Government on AI related matters.
- Specifically, on the Indian AI Stack, we believe there is need for broader participation from the government as the stack would address issues under the ambit of other ministries and deal with issues like cyber security, data protection, etc which are subjects regulated by other ministries.
- There is lack of clarity on the intent /purpose of the AI stack. It is unclear how an AI stack, which is general in nature and not tied to a specific application or a set of applications, would serve any meaningful purpose for the development of the AI ecosystem.
- The paper envisages a grand AI stack where everything ranging from the base data to the algorithm will need to be monitored and regulated. Therefore, the practicality of the proposal and the feasibility of developing and implementing such a system are important aspects that needs serious reconsideration.
- The efforts of the Committee to develop AI standards and an AI Stack for India, should not end up in creating India specific standards, which are not interoperable with the rest of the world. Alignment with global standards necessary.
- The paper makes a case for algorithmic audits and talks about the need for a horizontal AI regulator. These are idea floated with no adequate justification and therefore needs re evaluation.
- The paper suggests a open algorithmic framework. Though its unclear how this will pan out, we believe that open sourcing of algorithms should be voluntary.
- Also, there is no clarity of integration of AI stack with other govt led digital infra projects like the Health Stack.
The document s attached.