AI Experts Call for Safety Plan Amid Control Concerns

AI researchers warn of potential "catastrophic outcomes." 
 
A group of leading artificial intelligence (AI) scientists has issued a warning, calling on nations to establish a global oversight system to prevent potentially catastrophic consequences if control over AI technologies is lost. 
 
In their statement published on September 16, the team voiced concerns that the technology they have helped advance could pose severe threats should it fall outside human control. The statement reads: “Loss of human control or malicious use of these AI systems could lead to catastrophic outcomes for all humanity." It went further to highlight that we are yet unprepared with the necessary scientific tools and knowledge required to regulate and protect against misuse of such advanced intelligence. 
 
The scientists agreed that there's an urgent need for individual nations to create authorities capable of detecting and responding promptly to any AI-related incidents or risks within their jurisdictions. They also suggested developing a global contingency plan as part of long-term strategies. 
 
"In the longer term, states should develop an international governance regime to prevent development models that could pose global catastrophic risks," read part of the statement. 
 
This call comes in light of findings from early September’s International Dialogue on AI Safety held in Venice. This was its third meeting organized by US-based nonprofit research group Safe AI Forum. 
 
Gillian Hadfield, Professor at Johns Hopkins University who shared this statement online, asked rhetorically about what course would be taken if catastrophe struck due to autonomous self-improving models in six months' time—who would be called upon? 
 
These experts defined AI safety as being essential for worldwide public good, hence necessitating international cooperation and governance measures. 
  
They proposed three key initiatives: setting up emergency preparedness arrangements and institutions; creating a framework ensuring safety assurance; and carrying out independent universal research into AI safety verification. 
 
Related: Elon Musk voices support for sweeping regulations aimed at ensuring safe use 
 
Over thirty signatories endorsed this appeal, including some from Canada, China, Singapore, Britain, and the United States, among other countries. This group consists of experts from top AI research institutions and universities, including several Turing Award winners—the computing equivalent of a Nobel Prize. 
 
The scientists noted that this dialogue is increasingly necessary due to dwindling scientific exchange between superpowers and growing distrust primarily between China and the US—factors that complicate consensus on AI threats. 
 
Interestingly, in early September, the EU, UK, and US signed what's regarded as the world’s first legally binding international treaty on AI, which prioritizes human rights alongside accountability in regulating AI technologies. 
 
However, tech corporations and executives have voiced concerns over possible stifling of innovation due to overregulation, especially within European Union jurisdictions.