EU should be more rigorous in AI regulation: experts

AI News

Timnit Gebru in 2018.

Kimberly White | Photo Getty Images

A group of prominent artificial intelligence experts has called on European officials to pursue broader regulation of the technology in the European Union’s AI law.

In a policy brief released Thursday, more than 50 individual professional and institutional signatories called for the inclusion of General Purpose AI (GPAI) in upcoming regulations, rather than restricting regulations to a narrower definition. advocating for

This group, which includes institutions such as the Mozilla Foundation and experts such as Timnit Gebru, believes that general-purpose tools may not be designed with high-risk uses in mind, but can be used in a variety of settings that increase risk. said to be sexual. The group cites generative AI tools that have grown in popularity over the past few months, such as ChatGPT.

According to Mehtab Khan, signatory, resident fellow and leader of the Yale/Wikimedia Initiative, regulations on how AI is developed should be considered, including how data is collected and who was involved in collecting and training the technology. . About intermediaries and information.

“GPAI should be regulated throughout the product cycle, not just at the application layer,” Khan said, noting that simple high-risk and low-risk labeling “intrinsically fails to capture the dynamism of the technology.” added.

The group suggests that European policymakers take steps to future-proof the law, such as avoiding restricting rules to certain types of products such as chatbots. , warns that developers cannot avoid liability by pasting standard legal disclaimers.

Sarah Myers West, managing director of the AI ​​Now Institute, who spearheaded the policy brief, says the rise of mainstream generative AI tools like ChatGPT came after the EU’s early drafts.

“I think this kind of wave of attention to generative AI has raised awareness of this clause,” said Myers West. “But even before that, there was a wide variety of artificial intelligence that wasn’t tooled for a specific purpose, and would have gotten this kind of exemption as well.”

“As far as we know, EU AI is poised to become the first comprehensive regulation on artificial intelligence,” said Myers West. “And given that, it’s going to be a global precedent, and it’s especially important to do well with this category of AI because it could be a template for others to follow.” ”

Watch: How Nvidia grew from gaming to AI giant and now underpins ChatGPT

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *