Michelle Donelan, the technology secretary, aims to establish tech early warning signs through a safety summit
UK ministers have issued a warning, emphasizing the necessity of a “smoke alarm” system for dangerous AI, aiming to prevent severe threats like widespread casualties, cyberattacks, and uncontrolled AI technology. The technology secretary, Michelle Donelan, expressed hope that an upcoming safety summit in the UK would pave the way for an early warning mechanism. This mechanism would require tech companies to proactively identify risks in their AI products and be prepared to address them. Donelan stated, “We require a system akin to a smoke alarm in place. Companies should not only detect potential risks but also possess a plan to mitigate them. Such a comprehensive system should be implemented universally.
Addressing the upcoming two-day summit set to take place at Bletchley Park in November, Donelan emphasized the need to grasp the risks associated with AI, despite the incredible opportunities it offers. The government announced that the summit would primarily concentrate on two key areas: the misuse of AI for developing bioweapons or conducting cyber-attacks, and the challenge of controlling highly advanced AI systems.
Rishi Sunak reportedly believes that time is running short to establish a worldwide consensus on defining the most critical AI risks and devising strategies to address them. This urgency arises as technology companies leverage enhanced computing power, breakthroughs in technology, and increased investments to create increasingly potent AI models.
The summit is not anticipated to result in an international treaty akin to those for nuclear weapons regarding AI development. Instead, it is expected to delineate the spectrum of substantial risks posed by AI systems and suggest measures for their mitigation. The primary focus will be on “frontier” AI models, referring to state-of-the-art systems whose capabilities rival or surpass the most advanced existing models, potentially posing threats to human life.
The government stated that the summit aims to pinpoint the current state of AI development’s frontier and its potential trajectory. The event, held at Bletchley Park in Buckinghamshire, historically renowned as the workplace of codebreakers like Alan Turing during World War II, will convene global leaders, AI firms, academics, and civil society groups.
According to a government statement released on Monday, there is a concern that models significantly more powerful than the current ones, such as OpenAI’s ChatGPT powered by the GPT-4 model, could soon become available. These models possess capabilities that are extremely challenging to predict, even for those developing them, and there is a default risk of them being accessible to various entities, including those with malicious intent.
One of the primary worries among AI experts is the potential for advanced systems to surpass human control. This apprehension centers on the possibility of achieving artificial general intelligence, which refers to AI with intelligence levels equal to or exceeding that of humans, and the theoretical ability of such systems to circumvent any safeguards in place.
The government’s statement, providing a preview of the upcoming summit on Monday, highlighted the focus on systems that could evade control. It mentioned the risks associated with the “loss of control” stemming from advanced systems, emphasizing the importance of aligning these systems with our values and intentions.
