November 16, 2011 - Technology media outlet engadget published a blog post yesterday (November 15) thatEuropean Unionhas published on November 14th theGeneral artificial intelligencePreliminary draft Code of PracticeThe goal is to develop and use general-purpose artificial intelligence (GPAI) modeling the regulatory framework.
Generalized Artificial Intelligence Model Definition
The first draft of the Generic AI Code of Practice defines it as:
- Artificial intelligence models, including artificial intelligence models that are trained by self-supervision at scale using large amounts of data, that show remarkable versatility and are capable of a wide variety of different tasks, regardless of the manner in which the model is brought to market, and that can be integrated into a variety of downstream systems or applications.
Guidelines on systemic risk
Citing the blog post, the initial draft details how to manage the risks associated with generalized AI models, with guidelines that focus on helping organizations identify potential risks and take appropriate steps to mitigate those risks.
In addition the document provides specific recommendations for enterprises to comply with relevant laws and regulations, helping them to be able to effectively avoid legal risks when developing and applying AI technologies.
The EU is considering a tiered approach to regulation of generative AI models, with the higher the complexity and potential risk of the model, the more stringent the regulatory requirements. This tiered approach would help ensure that more powerful AI technologies are more closely regulated when they are used.