Google on Monday announced a $1 million grant to IIT Madras to set up a multi-disciplinary centre for responsible artificial intelligence to study and research various aspects of bias in AI while focusing on the Indian context.
AI and agriculture
The tech giant is also going to offer a similar amount to independent non-profit Wadhwani AI in order to develop AI models which will help monitor crop diseases, predict yield outcomes and bring efficiencies to Kisan call centers.
At the Google for India 2022 event, the company also announced that it will contribute to the agri stack as defined by the government of India through a combination of AI and remote sensing to better manage the problems arising in agriculture. Manish Gupta, Director of Google Research India, told the Indian Express that the tech giant will use its AI models over satellite images to identify things like the location of farms and farm boundaries to better understand the agricultural landscape.
Gupta also said that the plan is to apply additional analysis to understand the estimated yield, what crop is being grown where, which will help the financial institutions and policymakers.
The tech giant has already initiated a pilot project with the Telangana government and has also been working with a few startups in order to make the base level available for the solution.
Google plans to use imagery over a period of time to track changes in boundaries and detect activities like sowing, harvesting, and even stubble burning. The company also plans to onboard researchers working on forecasting floods and pest outbreaks to make use of all the capabilities.
Google’s project Vani
Google is also working on Project Vani, an AI model that attempts to better understand different languages, accents, and dialects. The company is working with partners like the Indian Institute of Science to collect speech data from all the districts of India to help cover the full linguistic diversity of India. This open source data will ensure that respondents can speak in any language that the users are comfortable with.
The company is improving its earlier model for Indian languages by including a hundred languages to help it understand text and speech.
Google to digitse handwritten medical prescriptions
The company's research team is also working to digitise handwritten medical prescriptions to capture the sitting information. Gupta said that though the technology is not new, it is difficult to extract things like a medicine's name from the medical prescription.
“There is so much information that is sitting in the pharmacists' heads that they used to decipher, and what we have been doing is, instead of using any rules, looking at how AI could capture some of that same kind of information to improve the accuracy of your handwriting recognition or OCR models,” Gupta added.
