Posted by AI has the potential to generate very important economic and social benefits. Concerns about the technology, however, have sparked demands for authorities to develop “responsible AI” without stifling research in the sector.
Ten principles have been outlined in a recent paper by Centre for Data Innovation Director Daniel Castro to aid policymakers in designing and analysing regulation approaches for AI that do not damage innovation. These are outlined here for HR professionals and companies’ convenience:
- Avoid pro-human biases: Allow AI systems to do what is legal for humans (while also prohibiting what is forbidden). Reasoning: Holding AI systems to a higher standard than humans disincentivises their adoption.
- Regulate performance, not process: Address AI safety, efficacy, and bias concerns by regulating results rather than developing explicit rules for the technology. Reasoning: Performance-based laws provide for greater flexibility in meeting objectives while avoiding the imposition of potentially costly and needless constraints on AI systems.
- Regulate sectors, not technologies: Set standards for specific AI applications in specific industries rather than broad rules for AI technology in general. Reasoning: The context is important. Even though they use similar underlying technology, an AI system used to drive a vehicle differs from one used to automate financial trades or diagnose illnesses.
- Avoid AI myopia: Address the entire problem rather than focusing on the AI component of a problem. Reasoning: Many problems must be tackled, whether or not they include AI. Focusing just on the AI component of the problem frequently diverts attention away from the larger issue.
- Define AI precisely: Define AI precisely to prevent mistakenly bringing other software and systems under the purview of new legislation. Reasoning: AI is integrated into numerous goods and spans a wide variety of technology. If policymakers merely seek to control machine learning or deep learning systems, they should avoid using broad definitions of AI.
Related link: 7 ChatGPT Hacks for University Students
6. Enforce existing rules: Hold AI responsible for following existing regulations. Reasoning: Many laws currently address common concerns regarding AI, such as workplace safety, product liability, discrimination, and others.
7. Ensure benefits outweigh costs: Take into account all of the potential costs and benefits of rules. Reasoning: Costs, including direct compliance costs as well as indirect innovation and competitiveness costs, have an impact on the merits of a regulatory proposal.
8. Optimise regulations: Maximise the benefits and minimise the costs of regulations.
Reasoning: Policymakers should find the most efficient way to achieve their regulatory objective.
9. Treat firms equally: Apply rules equally to firms regardless of their size or where they are domiciled. Reasoning: Exempting certain firms from regulations creates an uneven playing field and puts consumers at risk.
10. Seek expertise: Augment regulatory expertise with technical and industry expertise. Reasoning: Technical experts can help regulators understand the impact of regulatory options.