Roundup: March 2024
Some of the things we were up to last month..
Research
We have been investigating the readiness of LLMs for application to various domains in India. The pilot experiment on quantifying the deployability of this technology in the Indian legal context has made significant progress. Initial findings are published as a pre-print titled "InSaAF: Incorporating Safety through Accuracy and Fairness / Are LLMs ready for the Indian Legal Domain?” It is the result of a long-running collaboration between researchers at CeRAI and the PreCog group at IIIT Hyderabad. Read the preprint here.
Recent work on AI safety and RAI titled “Deception in Reinforced Autonomous Agents: The Unconventional Rabbit Hat Trick in Legislation” by Atharvan D (researcher at CeRAI) and team (of prominent researchers and experts in AI safety and legislation from Princeton, Stanford, Allen Institute of AI and Georgia Tech) is submitted and under review at the Conference of Language Modelling.
CeRAI in the media
Financial inclusion research covered by the Financial Express
Work on AI-based solution for digital financial inclusion of migrant workers, by CeRAI PIs Prof. Ravindran, Prof. Padmanabhan and their student, was covered by the Financial Express. This project was an impetus for founding CeRAI.
Manthan Summit by the Business Standard
Organized by a pre-eminent Indian business newspaper in its 50th year, Manthan was two-day summit on the major economic and business policy options and challenges that face the Indian economy and the many sectors that have the potential of driving the country’s economic growth.
Centre head Prof. Ravindran took part in a panel discussion on “How India can ride the AI wave to become a global powerhouse”
You can watch the recording from the 1:25:00 mark here:
Some of the key points discussed: How AI as it is today is too expensive for mass use in India, the importance of AI safety given its wide scope and its domain-specific nature, the need to rethink education in the age of AI, and the possibility of use of AI/VR in education.
Events
As part of IITM’s Open House (see coverage by The Hindu), CeRAI along with sibling centres under the Wadhwani School of Data Science and AI (WSAI) set up a booth for the general public to learn about AI through interactive games and illustrative posters. This was well received by visitors of all age groups and backgrounds.
Education
We announced three programs:
A summer school on RAI this June on IIT Madras campus, in collaboration with ACM India and Precog at IIIT Hyderabad.
*Responsible AI - Tour*, a 2-week executive certificate programme with the Centre for Outreach and Distance Education at IITM
*Data Science and Artificial Intelligence - Leadership Essentials*, a 2-month course providing an overview of DS/ML along with deployability and responsibility aspects of AI:
People
We were pleased to welcome to CeRAI Urvashi Aneja and Tarunima Prabhakar as Associate Research Fellows. Urvashi is a researcher, policy analyst and entrepreneur whose work seeks to drive an equitable and just distribution of technology gains. Tarunima is a researcher working at the intersection of technology, policy and global development.
Miscellaneous
We continue to hire actively. Apply here, or recommend friends, students and colleagues!
We have started blogging and tweeting more frequently. Please keep up to to date on our website, on Twitter, and of course here on Substack!