In an effort to confront the prevalent issue of inherent biases within generative artificial intelligence (AI) chatbots, major players such as Amazon and Microsoft are actively engaged in projects aimed at mitigating biases associated with large language models (LLMs). Rajeev Rastogi, Vice President of Machine Learning at Amazon, is at the forefront of these endeavors, leading projects that involve collaboration with Amazon India employees.

Furthering the cause, the Bill and Melinda Gates Foundation is supporting a three-year initiative by Bengaluru-based AI startup Karya. The project focuses on developing a methodology to ascertain, with a certain level of confidence, whether a given dataset exhibits gender bias. Additionally, Karya aims to establish a pipeline for the systematic debiasing of datasets. In response to inquiries directed at the Gates Foundation and Karya, Vivek Seshadri, Chief Technology Officer of Karya, revealed that the project is currently in its early research stages. Initial findings related to the mitigation of biases in language technology are anticipated to be available by mid-2024, at which point the company will be able to provide more comprehensive insights.

As the field of generative AI continues to evolve, these proactive initiatives underscore the commitment of industry leaders and non-profit organizations in addressing and rectifying biases within AI systems, ultimately striving for more equitable and unbiased technologies.