Sign In

Communications of the ACM

ACM Opinion

European AI Needs Strategic Leadership, Not Overregulation


View as: Print Mobile App Share: Send by email Share on reddit Share on StumbleUpon Share on Hacker News Share on Tweeter Share on Facebook
European AI Needs Strategic Leadership, Not Overregulation

"In the U.S, AI is being optimized to maximize corporate profitability and efficiency. In China, AI is being optimized to maximize the governments grip on the population with the preservation of power. The overly regulated environment in the EU will lead to complete chaos when regulations in various EU bodies start contradicting."

Mark Minevich is the principal founder of Going Global Ventures, a New York-based investment, technology, and strategic advisory firm.

With the global race to regulate AI officially on, the European Union (EU) Commission recently published a detailed proposal on how AI should be regulated, explicitly banning some uses and defining those it considers "high-risk," planning to ban the use of AI that threatens people's rights and safety.

The takeaways from the commission are incredibly in depth, but the ones that make the most sense to me are those that stress regulated AI should aim to increase human well-being. However, regulation should not overly constrain experimentation and development of AI systems.

From TechCrunch
View Full Article


 

No entries found