DeepMind CEO calls for an independent standards body to regulate frontier AI

Demis Hassabis, the CEO of DeepMind, has made a shocking proposal for an independent AI standards body that would oversee the development and release of frontier AI models, citing the need for rigorous testing and evaluation to prevent potential risks to society. The proposed body would be modeled after the Financial Industry Regulatory Authority, or FINRA, which regulates the financial industry in the United States. This move comes as AI technology is becoming increasingly powerful and ubiquitous, with many experts warning of the potential dangers of unregulated AI development. The proposal has significant implications for the tech industry, as it could lead to a major shift in the way AI is developed and released. For example, a recent survey found that 71% of AI researchers believe that AI development should be subject to stricter regulations, while 45% of respondents said that they had witnessed or experienced an AI-related incident that had significant consequences. The proposed standards body could help to mitigate these risks by establishing clear guidelines and best practices for AI development and release. Background context The idea of an AI standards body is not new, but Hassabis' proposal has brought it to the forefront of the conversation. In fact, the European Union has already established a high-level expert group on artificial intelligence, which has developed a set of guidelines for the development and use of AI. However, the EU's guidelines are voluntary, and many experts believe that a more robust regulatory framework is needed to ensure that AI is developed and used responsibly. For instance, a study by the MIT Initiative on the Digital Economy found that the development of AI could lead to significant job displacement, with up to 80% of jobs in some industries being automated. What to expect next The proposal for an AI standards body is likely to be met with significant debate and discussion, with some experts arguing that it could stifle innovation and others arguing that it is necessary to prevent potential risks. However, one thing is clear: the development and release of frontier AI models will require careful consideration and evaluation to ensure that they are safe and beneficial for society. The future of AI development The regulation of AI is a complex issue, and it will require a nuanced and multifaceted approach. For example, the city of New York has established a task force to explore the potential risks and benefits of AI, and to develop strategies for mitigating any negative consequences. As AI technology continues to evolve and improve, it is likely that we will see more proposals and initiatives aimed at regulating its development and use. The key takeaway from Hassabis' proposal is that the development of AI must be subject to rigorous testing and evaluation to prevent potential risks to society, and that an independent standards body could play a crucial role in this process.