With February 15 as the first deadline for submitting proposals to get government support to develop state-of-the-art foundational artificial intelligence models, a top official said that the government will have autonomy over the projects selected for support.
“The selected model will be monetised, but the government will retain full autonomy to use it for public good,” said Abhishek Singh, Additional Secretary, Ministry of Electronics & Information Technology.
While addressing a Q & A session regarding the invitation of proposals for India’s foundational AI model, the MeitY bureaucrat also informed that there can be a consortium of companies, individuals, researchers and experts but once the model gets selected it needs to be converted into one entity.
As there are a number of existing open-source AI models, “The proposed model can be built on existing open-source AI models but that should have the ability to convince our experts”, added Singh.
Using diverse and representative india datasets, incorporating robust mechanisms for fair outcomes and should address cross-sectoral india-specific challenges are some of the highlighting parameters mentioned in the session.
Regarding funding, Singh stated that it will be milestone-based, provided through grants, and facilitated via compute credits.
Data security remains a key concern for both users and the government. Addressing this, Singh assured that the model will comply with the DPDP Act.
The proposals will be considered on the basis of if the invitee has a good track record in development and deployment in AI research, it should have a multidisciplinary experience, it needs to have prior experience in building models, etc.
The government of India has envisioned India’s first AI model to be culturally and contextually relevant, which can address multiple issues existing in a number of sectors and can support creative and scalable use cases with measurable societal and economic impact.
Using diverse and representative india datasets, incorporating robust mechanisms for fair outcomes and should address cross-sectoral india-specific challenges are some of the highlighting parameters mentioned in the session.
Regarding funding, Singh stated that it will be milestone-based, provided through grants, and facilitated via compute credits.
Data security remains a key concern for both users and the government. Addressing this, Singh assured that the model will comply with the DPDP Act.
The proposals will be considered on the basis of if the invitee has a good track record in development and deployment in AI research, it should have a multidisciplinary experience, it needs to have prior experience in building models, etc.
The government of India has envisioned India’s first AI model to be culturally and contextually relevant, which can address multiple issues existing in a number of sectors and can support creative and scalable use cases with measurable societal and economic impact.
