Regulating AI: Finding the Balance between Innovation and Control
In a recent blog post titled “7 principles for getting AI regulation right,” Google has expressed its stance on AI regulation, emphasizing the need for a balanced approach. The company recognizes the importance of regulating AI to mitigate potential risks but also highlights the need to avoid stifling innovation. According to Kent Walker, president of global affairs for Google and Alphabet, the race in AI is not just about who invents something first, but about who deploys it effectively across various sectors.
Google, along with AI companies like OpenAI, has been vocal about the need for regulation due to the potential existential risks associated with AI. Google CEO Sundar Pichai even participated in the Senate’s AI Insight Forums to contribute to the discussion on legislating AI. However, there are concerns that Google and other AI companies are using fear-mongering tactics to gain regulatory capture and control over the industry. Yann LeCun, Chief AI Scientist at Meta, suggests that if these fear-mongering campaigns succeed, it could result in a situation where only a few companies have control over AI.
While Google supports the efforts of state and federal governments in AI regulation, it believes that legislation should focus on regulating specific outcomes rather than imposing broad strokes laws that hinder development. Walker references the White House AI executive order, the U.S. Senate’s AI policy roadmap, and recent AI bills in California and Connecticut. He argues that intervention should occur at points of actual harm and that blanket research inhibitors should be avoided. Walker notes that over 600 bills related to AI have been proposed in the U.S., highlighting the need for a more targeted approach.
The issue of copyright infringement and data usage in AI training models is also addressed in Google’s post. Companies using AI models argue that utilizing publicly available data constitutes fair use. However, they have faced accusations of copyright violation from media companies and record labels. Walker supports the fair use argument but also acknowledges the need for transparency and control over AI training data. He suggests that website owners should have the ability to opt out of having their content used for AI training through machine-readable tools.
One principle outlined by Google, “supporting responsible innovation,” touches on the issue of known risks in AI development. However, it does not delve into specific measures to prevent inaccuracies in generative AI responses that could lead to misinformation and harm. While the recommendation to put glue on a pizza may not have been taken seriously, it highlights the ongoing discussion surrounding accountability for AI-generated falsehoods and responsible deployment.
In conclusion, Google’s perspective on AI regulation emphasizes the importance of striking a balance between regulation and innovation. The company supports targeted regulation that addresses specific outcomes of AI development rather than inhibiting overall progress. It also acknowledges the need for transparency and control over AI training data while promoting responsible innovation. As governments navigate the complex landscape of AI regulation, finding this balance will be crucial to ensure the safe and beneficial deployment of AI technology.