As nations grapple with fast-moving AI technology and the best way to regulate it, could ethical human behavior be a preferable solution to government policy?

The European Union is on the brink of establishing the world’s first comprehensive legal framework for artificial intelligence, but not without its hurdles. Initially proposed in 2021, the EU AI Act has garnered attention from experts worldwide, with Stanford HAI hosting a discussion to dissect its intricacies. As highlighted by European Parliament Member Dragoș Tudorache, critical areas of contention include using AI for biometric surveillance, defining high-risk AI, and governance structures. Meanwhile, Rishi Bommasani from Stanford HAI identified gaps in compliance among major AI companies, such as OpenAI and DeepMind

As the EU forges ahead, the act’s implications extend beyond its borders, potentially influencing the U.S.’s approach to AI regulation. Does the U.S. need sweeping legislative action, or can the market police itself? Jon Stine, Executive Director at The Open Voice Network; raised concern over additional regulations when ethical human behavior can make AI work for the betterment, not a detriment.

Jon’s Thoughts

“I think we have a wonderful, very complex set of privacy regulations. One might wish we would have a federal privacy regulation, but the current quilt of state regulations, California, Illinois, et cetera, give us an overall guidance to respect and protect data. You add that to the European General Data Protection Regulation, GDPR, you have a guide on how to respect and how to protect data. I don’t know that we need more laws protecting privacy. We need more observance of existing law. And maybe we need to, you know, rough off the different edges between state laws, sure. You’d be hard-pressed to have someone say, in the enterprise world, who’s giving it any bit of thought, that we don’t know what the privacy law is. No, there’s a lot of privacy law.

In AI, this is moving so fast. Yes, there is the European AI Artificial Intelligence Act coming in April. That will give us guidance, much as GDPR gave us guidance. I’ve read the act. It’s not an easy read. There’s good general guidance. But there’s also, we have a whole body of ethical AI literature out there. There are libraries, books, articles, documents. And we read that, and we realize that the AI Act, which is a law, and then just general ethical principles about the use of AI, about the need for accountability, consent, and transparency. These are not recent ideas. This is stuff 10, 15, 20, 30 years ago, which all basically demands we as humans pause, ask what’s being done, who it’s being done with or to who, and be thinking about potential harms. And then taking responsibility for what is done. I don’t think we need more laws to do that. We simply need humans to take responsibility and behave that way. Maybe it takes a law to change a human. Someone will have to tell me that.”

Article by James Kent