ABSTRACT
Sustainable Development Goal 4 (‘SDG 4’) which aims to ‘ensure inclusive and equitable quality education and promote lifelong opportunities for all’. The Data Dilemma: How AI Models Learn AI is frequently discussed alongside machine learning (‘ML’) and deep learning (‘DL’), which are disciplines integral to AI algorithm development. These algorithms are modeled after the decision-making processes of the human brain. The more extensive and diverse the dataset fed into ML and DL systems, the more proficient the system becomes at emulating human cognition. ChatGPT is a Large Language Model (‘LLM’), a sophisticated subset of AI, ML, and DL technologies, trained on vast amounts of data available on the internet to excel at ‘next word prediction’ while simultaneously learning grammar, syntax, and acquiring comprehensive knowledge about the world. The role of AI in ensuring achieving the SDG4 cannot be ignored however the training of AI raises several questions which are pending before the judicial courts around the world (including India).
AI models are trained on extensive datasets sourced from the internet, which frequently include copyrighted materials without transparent disclosure regarding their origins. This opacity has triggered several class-action lawsuits against major technology companies like Microsoft, Meta, and emerging entities such as OpenAI, alleging copyright infringement during the AI training process. This point of contention is recently raised in India before the Delhi High Court, wherein ANI Media has filed a suit against OpenAI (in which additionally, Digital News Publishers Association and Indian Music Industry have filed to join the suit) for using former’s large archive comprising text, exclusive statements, articles, programmes, interviews, video content, images/photographic content, etc.
Singh, Sapna, Feeding The Algorithm: Legal Challenges In AI Training Data (February 21, 2025), Journal of Global Studies, Legal Studies (Vol 2: LS), ISSN Number 2393-9354.
Leave a Reply