Artificial intelligence (AI) is a powerful technology that can bring great benefits to humanity, but also pose serious challenges and threats. Many people are concerned about the potential dangers of AI in the future, such as superintelligent machines that could outsmart and harm humans.
However, according to a new article by Stuart Russell, a professor of computer science at the University of California, Berkeley, and Allan Dafoe, a professor of international politics at the University of Oxford, we should not neglect the AI risks that are already present today. These include cyberattacks, misinformation, surveillance, discrimination, manipulation and autonomous weapons.
The authors argue that these risks are not hypothetical or speculative, but real and urgent. They call for more regulation and governance of AI to ensure its safety and alignment with human values. They also propose a framework for assessing and mitigating AI risks based on four dimensions: severity, likelihood, imminence and tractability.
They suggest that regulators should prioritize the most severe and likely risks that are imminent and tractable, such as cyberattacks and misinformation. They also recommend that researchers and developers should adopt responsible AI practices, such as transparency, accountability and ethics. Moreover, they urge the international community to cooperate and coordinate on AI governance, especially on issues that affect global security and stability.
The authors conclude that by addressing the AI risks right in front of us today, we can better prepare for the future challenges and opportunities of AI. They say that this is not only a moral duty, but also a strategic necessity.
