Background:

• Artificial Intelligence is the field of developing computers and robots that are capable of behaving in ways that mimic human capabilities and sometimes go beyond it. 
  • AI-enabled programs can analyse data to provide information or automatically start actions without human interference.
• Machine learning is a pathway to artificial intelligence. This subcategory of AI uses algorithms to automatically learn insights and recognize patterns from data, applying that learning to make increasingly better decisions.

Perils in the use of AI & ML:

• AI Bias: AI models are becoming prone to biases, often due to skewed or unrepresentative training data, which can lead to discriminatory outcomes.
• Misinformation: AI systems can propagate misinformation, either unintentionally through flawed models or deliberately through data manipulation. This can be a problem during critical events like elections.
• Privacy Breaches: The use of sensitive data in AI models heightens the risk of privacy breaches, with unauthorised access or misuse of personal information.
• Data Poisoning: There is a risk of data poisoning, where malicious actors deliberately introduce manipulated data into AI systems, leading to biassed or adversarial outputs.
• Data Management Challenges: The continuous processing and churning of vast amounts of data create challenges in effectively correcting, updating, and erasing sensitive data from AI models.
  • The complex nature of ML models, like the Large Language Models (LLMs) and deep neural networks, makes it difficult to manage and control the data processed by these systems.
• High Computational Costs: Addressing issues like data errors or biases through methods such as retraining AI models can lead to inflated computational costs and delays.
• Lack of Regulation and Standards: The absence of clear regulatory frameworks or uniform standards for AI use across different jurisdictions creates challenges in ensuring ethical and responsible AI deployment.
• Geopolitical and Transboundary Implications: The global nature of AI development and its transboundary implications make it challenging to establish uniform governance and regulatory standards across different countries.

Measures taken in India to regulate AI & ML:

• The NITI Aayog released the National Strategy for Artificial Intelligence, which featured AI research and development guidelines focused on healthcare, agriculture, education, “smart” cities and infrastructure, and smart mobility and transformation.
• The Digital Personal Data Protection Act,2023 which is India’s data privacy law can be leveraged to address some of the privacy concerns concerning AI platforms.
• India is a member of the Global Partnership on Artificial Intelligence (GPAI). This partnership aims to work “as a vital branch of the initiative, GPAI’s Experts produce deliverables that can be integrated into Members’ national strategies to ensure the inclusive and sustainable development of AI. 

Measures needed to regulate AI & ML:

• Development of Machine Unlearning (MUL) Techniques: Develop and implement Machine Unlearning algorithms to allow AI models to effectively delete false, outdated, or sensitive data, thereby improving data management and reducing biases.
• Promotion of Ethical AI Practices: Encourage companies to adopt ethical AI practices, including transparency in AI decision-making processes and regular audits of AI models to detect and mitigate biases and other issues.
• Enhanced Regulatory Frameworks: Governments should create and enforce clear regulatory guidelines for AI usage, including provisions for data privacy, bias reduction, and misinformation control, similar to the EU’s AI Act addressing data poisoning.
• Investment in AI Research and Development: Increase investments in AI R&D to develop more robust, fair, and secure AI models, particularly focusing on reducing biases and improving data handling capabilities.

Conclusion:

• Different stakeholders such as the Governments, Corporates and People should come together to address technical and regulatory considerations to ensure effective implementation of AI & ML in this evolving landscape.