UK and US Sign Historic Agreement on AI Safety

The UK and US have agreed to collaborate on testing advanced artificial intelligence (AI), marking a significant milestone in their partnership.

An agreement was signed on Monday, highlighting the collaboration between both countries in the development of reliable methods for assessing the safety of AI tools and their underlying systems.

This marks a significant milestone as the first bilateral agreement of its kind.

According to UK tech minister Michelle Donelan, this is a significant technological challenge that we must address.

“It has been consistently emphasized that the safe development of AI is a matter of global concern,” she stated.

“It is crucial that we collaborate in order to confront the challenges posed by this technology and utilize its vast potential to improve our quality of life and well-being.”

The secretary of state for science, innovation and technology emphasized that the agreement further strengthens the commitments made at the AI Safety Summit held in Bletchley Park in November 2023.

At the event, prominent figures in the AI industry such as Sam Altman from OpenAI, Demis Hassabis from Google DeepMind, and tech billionaire Elon Musk came together. They announced the establishment of AI Safety Institutes in both the UK and US. These institutes will focus on assessing AI systems, whether they are open-source or closed-source.

Concerns about artificial intelligence

At present, most AI systems can only handle individual tasks that would typically require human intervention.

Referred to as “narrow” AI, these tasks can vary from swiftly analyzing data to delivering a desired response to a prompt.

However, there are concerns about the potential risks posed by advanced “general” AI tools that have the ability to perform a wide variety of tasks typically done by humans, which could potentially pose a threat to humanity.

“AI, similar to other scientific disciplines such as chemistry, nuclear physics, and biology, has the potential to be utilized as a tool for both positive and negative purposes,” Prof Sir Nigel Shadbolt shared with the BBC’s Today programme.

However, the University of Oxford professor mentioned that concerns regarding the existential risk of AI can be exaggerated at times.

“It is crucial that we show our support and gratitude for the initiatives taken by prominent AI experts to explore and analyze potential risks,” he stated.

“It is important to gain a comprehensive understanding of the vulnerability and capabilities of these models.”

Gina Raimondo, the US commerce secretary, highlighted the importance of the agreement in enhancing the governments’ comprehension of AI systems, enabling them to provide more effective guidance.

“It will enhance the progress of our Institutes’ efforts across a wide range of challenges, be it related to our country’s security or our society as a whole,” she stated.

“Our collaboration demonstrates our commitment to addressing these concerns head-on, rather than avoiding them.”