In July, the Council of Europe’s Committee on AI presented a consolidated working draft of the Framework Convention on Artificial Intelligence, Human Rights, Democracy, and the Rule of Law. This Draft is intended to be the basis for future negotiations for signing the definitive Framework Convention on AI, HR, Democracy, and the Rule of Law.
The proposal for article 1 of the Draft, which establishes purpose and object, provides a clear overview of the goals of the future Convention: “This Convention sets out principles and obligations aimed at ensuring that design, development, use and decommissioning of artificial intelligence systems are fully consistent with respect for human dignity and individual autonomy, human rights and fundamental freedoms, the functioning of democracy and the observance of the rule of law.”
- Some interesting remarks about this Draft are the following: It follows a risk-based approach (art. 2), meaning that risks should be mitigated, and measures should be taken according to the risks. This can be also seen in Transparency and oversight principle (art. 7).
- Definition of AI systems: “…any algorithmic system or a combination of such systems that uses computational methods derived from statistics or other mathematical techniques and that generates text, sound, image or other content or either assists or replaces human decision-making”. This is a somewhat vague definition. While content generation is evident, systems that “assist” or “replace” human decision making can be comprehensive.
- AI development and use should comply with human rights and fundamental values and not interfere with democracies and the rule of law. Non-discrimination is another key principle.
- Data protection requires that each party ensures privacy from design to decommissioning. It recognizes the importance of law but also standards on AI.
- It places obligations on parties to ensure the safety, security, and robustness of the functioning of AI systems. Importantly, it allows parties to create environments for testing AI systems for research and innovation.
- The Draft requires also the parties to implement safeguards and mitigation measures such as:
- Provision of information to affected individuals, including not only knowing that they are interacting with AI systems, but also information regarding the logic and consequences should be provided so they can contest the automated decision.
- As part of the risk management framework, take measures to identify, evaluate, and mitigate risks to human rights, democracy, and the rule of law. These impact assessments should also be consulted with the relevant stakeholders and be made publicly available, where appropriate.
- Impose bans or moratoriums on specific AI systems incompatible with fundamental rights.
- Provide training to those responsible for developing and using AI systems to enable them to identify, evaluate and mitigate risks to fundamental rights.
This Draft Convention establishes the principles and standards for the future regulation of AI, which will be vital to ensuring respect for human rights, non-discrimination, democracy, and the rule of law. The AI4Gov Project is perfectly aligned with the Draft and the future Convention since it aims to provide technical tools, protocols, standards, and training for implementing AI in democratic governmental decision-making as well as to educate the citizens and generate trust in data collection. AI4Gov encompasses several goals oriented to protect vulnerable groups such as women, and people with disabilities, among others, in line with the provisions of this Draft. AI4Gov will ensure that AI will help improve Democracy, Human Rights, and the Rule of Law and assist public officials and citizens to understand and avoid some potential pitfalls of this technology.