-
AuthorPosts
-
As artificial intelligence systems continue to play a growing role in various aspects of our lives, establishing legal and regulatory frameworks to ensure transparency and explainability has become a pressing need. Several jurisdictions are working on legislation to address these concerns.
European Union’s AI Regulation: The EU is in the process of developing a comprehensive legal framework for AI that aims to guarantee transparency, explainability, and ethical use of AI systems. The regulation, currently under discussion, emphasizes the following key aspects:
Risk-based approach: This proposed regulation classifies AI systems according to the level of risk they present to fundamental rights, imposing stricter requirements on high-risk systems.
Transparency requirements: AI systems must offer users information about their capabilities, limitations, and intended purposes.
Explainability requirements: High-risk AI systems need to be designed in such a way that users can comprehend and interpret the system’s outputs, while developers must supply documentation outlining the system’s inner workings.
Accountability and liability: The regulation puts forth a well-defined legal framework for liability, ensuring that companies deploying AI systems are held responsible for any harm resulting from the system.
Algorithmic Accountability Act in the United States: The proposed Algorithmic Accountability Act in the US aims to hold companies accountable for their AI systems, particularly those that could have discriminatory effects. The primary components of the proposed act are:
Impact assessments: Companies would be required to carry out regular evaluations of their AI systems, examining their effects on fairness, accuracy, bias, and privacy.
Transparency and public disclosure: The act would compel companies to share information about their AI systems, including the data types used, the system’s purpose, and the techniques involved in its design and training.
Remediation: In instances where an AI system is found to have undesirable impacts, the act would require companies to address and mitigate these issues.
- This topic was modified 1 year, 5 months ago by AI Standards Hub.
- This topic was modified 1 year, 5 months ago by AI Standards Hub.
-
AuthorPosts