Unveiling the projects Wednesday, as part of the Rs 10,738 crore IndiaAI Mission that comprises both monetary and non-monetary support activities to steer ethical AI in India
Selected through an Expression of Interest (EoI) under the mission's "Safe and Trusted AI" pillar, the 73 projects were picked from more than 2,000 proposals.
The government's AI initiative is structured along verticals, with one such theme focusing on bias mitigation, which cuts across five areas that also include privacy and governance of our work with artificial intelligence.
 |
| Photo by Tara Winstead from Pexels |
Other pillars such as the building of a 10+ PetaFlops AI supercomputer or an ecosystem around high-level computing boasting over 10,000 GPUs and also in crafting front-line recommendations for India's artificial intelligence models shapes up contributing to this vision.
IIT Jodhpur: IIT to work on 'machine unlearning' in generative foundation models Machine Unlearning is a method which enables a model to “forget” or remove some of the training data points without needing to parse through and retrain from scratch.
While IIT Roorkee will be dealing with 'generating synthetic data to mitigate bias in datasets', NIT Raipur's focus area for the study is developing strategies to overcome bias from healthcare systems.
A framework for the "development of validation dataset for facial recognition systems" will be jointly developed by CSIR-NPL New Delhi with DIAT Pune along-with Mindgraph Technology, Hyderabad and another joint initiative between CSIO Chandigarh and NILERD Goa.- TNS
In parallel, IIT Delhi has partnered with IIT Dharwad, IIIT Delhi and the Telecommunication Engineering Centre (TEC) to generate privacy-preserving machine learning models.
IIIT Delhi (Indraprastha Institute of Information Technology) working with TEC will be preparing tools to assess the fairness of AI models in a project called Nishpaksh.
Civic Data Labs has been awarded a grant to build ParakhAI, an open-source algorithm auditing toolkit. The collaboration between Amrita Vishwa Vidyapeetham and TEC aims at working on the ‘Track Low Level Metrics (LLM)’ framework that provides transparency and risk assessment of large language models.
The government also cleared the IndiaAI Mission worth Rs 10,732 crore in March which will produce computing infrastructure across different parts of the country, many AI centres devoted to build multi-modal LLMs and procure over 10,000 GPUs amongst other dedicated resources.