Stepping up AI Regulation: A Comparative Glimpse on US and EU Approaches

JJohn November 28, 2023 1:02 PM

As the world focuses on advanced AI research, the US government makes strides in regulating everyday AI systems. In contrast, the EU appears to lag behind. The article delves into this disparity, highlighting the US Office of Management and Budget's instrumental role and the challenges the EU faces in formulating effective AI regulations.

OMB's memo: A potential game-changer

The OMB, a unit within the US president's executive office, plays a pivotal role in the country's AI landscape. Its recent memo, released by director Shalanda Young, outlines a proposed policy that could significantly transform the usage of AI in the government sector. Although still in draft form, the proposal stipulates every department to designate a chief AI officer who would maintain a register of existing AI applications. These steps mark an important move towards greater transparency.

Balancing AI benefits against individual rights

The OMB's proposed policy doesn't stop at transparency. It further mandates closer examination of AI systems with potential implications on people's safety or rights. This means that the agencies will be required to evaluate the risks versus the purported benefits of these AI systems. They'll also have to verify the data's quality that they use and monitor the deployed systems more closely. This policy outlines that affected individuals should receive clear explanations of AI usage and should have an opportunity to challenge AI decisions.

The memo also tackles an under-discussed but significant aspect — government procurement of AI. It highlights the issue of inexperienced government agencies adopting new AI software that they don't fully understand, often resulting in severe consequences for the most vulnerable. The proposed policy recommends best practices for procurement of AI systems, thereby addressing these concerns. Furthermore, it emphasizes the need for consultation with affected groups before deploying such systems.

Contrasted with the US’s structured approach, the EU's attempts at regulating AI have been fraught with complications. These range from disagreements among member states to the challenge of keeping up with rapidly advancing AI research. The EU Parliament's attempt to add regulations for frontier AI systems to its existing AI Act met with resistance from France and Germany, who were concerned about potential impacts on their burgeoning domestic AI companies. The fast-paced AI ecosystem and the lack of sufficient public interest research add further complexities.

Lessons from the US for the EU

The EU could learn from the US’s approach, especially in terms of transparency and public discourse. Increased transparency from leading AI labs would be a good start, especially regarding any potentially dangerous capabilities of new AI systems. But beyond that, it would be wise not to rush policy development. Instead, fostering civil society debate and in-depth research, much like the process that informed the OMB memo, could lead to more robust and well-grounded policy proposals.

More articles

Also read

Here are some interesting articles on other sites from our network.