ECP | Platform for the Information Society Netherlands gladly contributes to the European Community of Best Practices to achieve robust, legal and ethical AI. For this purpose, we developed the AI Impact Assessment: a first of its kind practical tool that helps an organization to design, employ and audit AI (including the input and output data) in an ethical and legal manner.
Implementation blueprints per industry
Currently, our multidisciplinary teams are creating AIIA implementation blueprints per industry (Health, AgriFood, Energy, Finance, Transportation, Education, Media, Art & Entertainment etc) which can be used as an example for sector specific AIIA application. Additionally, we will present our guide to leading Silicon Valley companies in the United States this summer. As the HLEG Ethics Guidelines of April 8th 2019, the AIIA will also be updated on the basis of enhanced insights, audits, red teaming, use cases and feedback.
Safe implementation of AI
The AI Impact Assessment offers entrepreneurs, data scientists and software programmers a concrete code of conduct with which AI can be safely implemented. The Artificial Intelligence Impact Assessment was developed by the Dutch ECP | Platform for the Information Society. The AI Impact Assessment is a guide for the application of artificial intelligence. Through a practical checklist from a legal, technical and ethical point of view.
Code of Conduct Artificial Intelligence
The starting point of the AI Impact Assessment is the Code of Conduct for Artificial Intelligence. This code of conduct forms the basis of the AI Impact Assessment framework. The Artificial Intelligence Code of Conduct consists of two parts: common European ethical and constitutional values (i.e. 1791 liberty, equality, fraternity), legal principles (fairness, proportionality, rule of law) and democratic preconditions, as well as practical rules and codes of conduct for (inter alia) AI applications, the training data corpora, inference systems, deep learning algorithms, neural networks and autonomous systems.
Risk and safety assessment tools
The AI Impact Assessment is related to other risk and safety assessment tools such as the Privacy Impact Assessment, Algorithmic Accountability and Responsibility & Transparency by Design and the 23 Asilomar AI principles. The AI Impact Assessment also has a guide function, and provides a similar moral compass to achieve reliable, Trustworthy AI.
Guidance for responsible implementation of artificial intelligence
The AI Impact Assessment contains a comprehensive step-by-step plan with 8 steps, on the basis of which companies, developers and designers can clarify which legal and ethical standards are important in the creation and application of AI systems and smart robotics. The AI Impact Assessment and the Artificial Intelligence Code of Conduct also clarify which motives and considerations form the basis for the choices and decisions to be made during the AI implementation process. This makes the use of artificial intelligence, (fair and trusted) data and smart algorithms more transparent, as well when AI is deployed in conjunction with other 4th industrial revolution technologies such as DLT, Mixed Reality and Quantum Computing, to achieve synergetic effects.
Remove obstacles and roadblocks for AI pilots & start-ups
Finally, our AI Impact Assessment’s systemic and holistic approach aims to remove obstacles and roadblocks (such as intellectual property, privacy, sharing and pooling of quality datasets) for AI pilots, start-ups and scale-ups.
Download the English version of our roadmap to robust, legal and ethical AI here: https://airecht.nl/s/Artificial-Intelligence-Impact-Assessment-English.pdf