The legislation also addresses the issue of algorithmic bias, creating a new playbook for responsibly using AI based on inherently biased algorithms. It also bans specific technologies or use cases, shifting the debate to how technology companies can rebuild trust and combat biases. The impact of this legislation is expected to be minimal across the Department of Defense, the intelligence community and broader government sectors, and could potentially lead to the development of more trustworthy systems.
Key takeaways:
- The EU's AI Act legislation focuses on banning AI systems that rely on machine learning, logic and knowledge-based approaches that present "unacceptable risk" and "high risk".
- The legislation aims to rebalance equity by focusing on privacy and trust, but there are concerns that it may stifle innovation and limit the potential of dual-use technologies.
- The legislation does not challenge or debate issues of bias in AI, but instead creates a new playbook for how to responsibly use AI based on algorithms that are inherently biased.
- Compliance with the EU’s AI Act could provide the U.S. with more trustworthy systems, underscoring the existing mission to protect the privacy and rights of citizens, and could prompt a rethinking of internal processes and control systems.