The AI Safety Summit: Avoiding a Tech-Catastrophe

The three key emerging points from the AI summit are:

The Bletchley Declaration 

28 countries (including heavyweights such as the USA, the EU and China) have signed this aptly named declaration with a nod to the infamous wartime Codebreakers.

The declaration establishes a shared understanding of the opportunities and risks posed by AI development and the need to collectively manage potential risks through a joint global effort to ensure AI is developed and deployed in a safe and responsible manner.

In theory, such a joined-up approach is welcomed. Still, on a practical level, there will need to be an equal investment in the training of experts in adjudicating and enforcement.

It remains to be seen as to whether this be a global body or whether countries are left to implement their own systems, which doesn’t feel like it would sit with the global nature of AI that has no physical or jurisdictional boundaries at the moment.

It therefore raises the question as to whether we need gatekeepers of the gatekeepers, a two-tiered approach?

Safety Institutes

1. The AI Safety Institute - The UK announced its new AI Safety Institute, whose mission is to ‘minimise surprise to the UK and humanity from rapid and unexpected advances in AI’. The Safety Institute shall be tasked with testing the safety of new and emerging types of AI and assessing the potentially harmful capabilities.
2. U.S. Safety Institute - Vice President of the US, Kamala Harris, announced that the US is establishing an AI Safety Institute, which will develop technical guidance and enable information sharing and research collaboration with peer institutions, including the UK’s AI Safety Institute.
3. European AI Office - Similarly, the president of the EU Commission announced the launch of its proposed European AI Office, which shall delve into the most advanced AI systems. It is envisioned that the EU AI Office will have a global reach, helping to foster standards and testing practices for AI systems, and shall also cooperate with similar entities around the world, including the newly formed UK and US AI institutes.  

Isambard-AI 

The UK government announced a major £225 million investment into a new supercomputer called Isambard-AI, which will be built at the University of Bristol and is intended to achieve breakthroughs across healthcare, energy, climate modelling and other fields and forms part of the UK’s aim to lead in AI while partnering with allies like the US.

The Ministerial Foreword to the Government’s recently published policy paper on the AI Safety Institute, Rt Hon Michelle Donelan MP stated, “we were surprised by rapid and unexpected progress in a technology of our own creation”.

Really? Do we need to revisit some of the Hollywood blockbusters that have been warning us for decades?

Films such as:

• 2001: A Space Odyssey (1964) – a cult classic depicting an early interpretation of what could happen when AI rebels. 
• The Terminator (1984 onwards) – SKYNET - do we need to say anymore?·       
• Robocop (1987) - despite being over three decades old, it’s still thought of as a thought-provoking look into the ethics of AI and the dangers of misusing technology.
• The Matrix (1999) – we now have the ever-evolving Metaverse.
• Minority Report (2002) - predicting the not-so-distant future – the AI central to the film is a crime prediction tool. Similar A.I. tools have been deployed within various agencies in the U.S. since 2012.
• I, Robot (2004) – rogue self-thinking robots capable of human crimes – how would we begin to police this – send in Robocop?
• Chappie (2015)- highlights the perils of AI in criminal hands.

And the list goes on. Whilst Sci-fi films allow for exaggeration and imagination, it’s not beyond the realms of possibility that some fiction could become fact!

Do we need to book the film nights, revisit some classics, and start taking notes from these movies?