The prestigious Indian Institute of Technology, Madras (IIT-M) in association with two firms has launched an initiative called ‘Hidden Voices’ to reduce the gender gap in digital sources by setting a goal of auto-generating biographies of notable women within a year.
The IIT-M is partnering with Robert Bosch Centre for Data Science and Artificial Intelligence (RBCDSAI) and SuperBloom Studios, a business consultancy firm founded by three IIT-M alumnae, for this purpose.
The institute said in a statement on Friday that the founders of the initiative have set a goal of auto-generating biographies of several notable women within the next year, specifically within the next International Women’s Day (8th March 2023) to make a positive impact on gender representation among digital sources.
Hidden Voices is planned to start with generating biographies of women from North America and India who have notable contributions in STEM fields and tech-adjacent business domains. The team aims to expand expertise areas, geography and include other underrepresented communities over time.
It also said natural language models are increasingly forming the basis of various consumer interaction services and the models depend on open web datasets, including Wikipedia.
While there are multiple layers of complexity to resolve the nature of equitable representation across all digital platforms, it is noted that there is significant value in increasing women's representation in Wikipedia.
“The project will be an instance of a human-allied AI execution. While state-of-the-art of Automated Language Processing has significantly advanced there are situations when the AI will make errors. This is especially so when processing documents about underrepresented populations, the very fact that this project is trying to address,” Prof Balaraman Ravindran, Head, RBCDSAI-IIT Madras, said.
He added that they will take advantage of AI solutions where possible, and judiciously use “human oversight and verification to produce high-quality outputs.”
Dr Raji Baskaran, founding partner, SuperBloom Studios, said the lack of availability of proper information often creates and cements unintended biases.
“This is nowhere more prominent than in the ever-widening digital gender data gap. Hidden Voices addresses a critical data gap and builds tools to systematically reduce this gap at scale. Building products and services that are inclusive is at the core of our business strategy,” Dr Raji Baskaran said.
Some of the major barriers in addressing the data gap include editors' gender and interest but also contributions from external sources. Hence, the project aims to develop information theoretical approaches, ML-assisted auto-identification and validation of external sources and textual analysis methods to auto-generate the first draft of Wikipedia-style biography, the institute said.
Watch the latest DH Videos here: