A naukri.com initiative
Analyticsindiamag
1d
63
Image Credit: Analyticsindiamag
Open Source LLMs Pave the Way for Responsible AI in India
- Open-source large language models are being used in India for responsible AI by fine-tuning models on locally relevant datasets.
- Projects in healthcare, agriculture, and education leverage LLMs supported by Meta, offering freedom in model customization.
- Initiatives like oral reading fluency assessments in Gujarat use AI4Bharat's open-source models, combining student data for training.
- Banerjee emphasized post-training on existing pre-trained models for vertical use cases, promoting public domain practices.
- Open source does not inherently eliminate bias; fine-tuning datasets is crucial, with transparency enabling public scrutiny to identify errors.
- Positive bias can be introduced to improve accuracy in AI models, balancing inherent limitations and specific domain requirements.
- Security concerns in social sector organizations focus on PII protection, with ethical governance principles requiring domain-specific guidelines.
- Banerjee advocates for an inter-governmental initiative for AI safety resembling aviation safety, addressing ethical standards in AI development.
- India's emphasis on AI for social good is supported by government collaboration, with open-source AI contributing to solving challenges across various sectors.
- The use of open-source LLMs reflects India's progress in responsible AI, fostering innovation to address critical societal issues effectively.
Read Full Article
3 Likes
For uninterrupted reading, download the app