It complements past publications such as: (1) the Model AI Governance Framework ("Model Framework"), which provides detailed and ready-to-implement guidance to private sector organizations on how to address key ethical and governance issues when deploying AI solutions; (2) its accompanying Implementation and Self-Assessment Guide to help organizations assess the alignment of their internal processes with the Model Framework; and (3) a Compendium of Use cases to showcase the successful implementation of Model Framework measures.
Organizations would do well to participate in piloting the MVP in order to gain early access to its self-testing capabilities, use MVP-generated reports to demonstrate transparency, and help develop an MVP that accurately reflects industry needs and contributes to the development of international standards.
The Pilot Testing Framework and Toolkit does not, however, guarantee that any Al system tested under this Pilot Framework will be free from risks or biases or is completely safe.
In more detail
On 25 May 2022, Singapore launched AI Verify, the World's first AI Governance Testing Framework and Toolkit, which is intended for companies wishing to demonstrate responsible AI in an objective and verifiable manner. This is currently available as an MVP for system developers and owners who want to be more transparent about the performance of their AI through a combination of technical tests and process checks.
Broadly, developers and owners can verify the claimed performance of their AI systems against a set of principles through standardized tests. Self-assessment is carried out using a toolkit that comprises a package of open-source testing solutions and is capable of generating reports for developers, management, and business partners, covering major areas affecting AI performance. The areas covered are aligned with desiderata set out in internationally accepted AI ethics, principles, guidelines and frameworks. These include, but are not limited to:
- Transparency:
- On the use of AI to achieve what stated outcome
- Understanding how the AI model reaches a decision
- Whether the decisions predicted by the AI show unintended bias
- Safety and resilience of AI system.
- Accountability and oversight of AI systems.
As it is in the nascent stages of development, the MVP currently:
- Works with a certain subset of common AI models but does not currently support unsupervised models;
- Can handle tabular datasets but has limited support for image datasets; and
- Supports small-to-medium scale models (around two GB) that can be fully imported using a web interface.
Related articles
Singapore: The High Court issues injunction to block potential sale and transfer of NFT
Singapore: PDPC publishes guide on responsible use of biometric data in security applications
Singapore: Government proposes Codes of Practice to regulate harmful online content on social media
© 2022 Baker & McKenzie.Wong & Leow. All rights reserved. Baker & McKenzie.Wong & Leow is incorporated with limited liability and is a member firm of Baker & McKenzie International, a global law firm with member law firms around the world. In accordance with the common terminology used in professional service organizations, reference to a "principal" means a person who is a partner, or equivalent, in such a law firm. Similarly, reference to an "office" means an office of any such law firm. This may qualify as "Attorney Advertising" requiring notice in some jurisdictions. Prior results do not guarantee a similar outcome.
