AI Safety Summit 2023 - What Have We Learned?

• Content
• Smart Buildings & Technology
• AI Safety Summit 2023 - What Have We Learned?

13 November 2023

Representatives from 28 countries recently met at Bletchley Park in Milton Keynes to discuss the risks and opportunities around AI.

 

Watch the Video

 

 

Let’s review the main takeaways from the AI Safety Summit 2023:

 

The Bletchley Declaration Was Signed

 

A declaration which establishes a shared understanding of the opportunities and risks posed by frontier AI was signed by 28 countries from across the globe. The signatories agreed to share knowledge on AI safety and research, as well as intelligence about AI’s misuse.

Prime Minister Rishi Sunak called it “a landmark achievement that sees the world’s greatest AI powers agree on the urgency behind understanding the risks of AI.”

 

Picture: a graphic from the summit which says: "AI Safety Summit - Hosted by the UK 1-2 November 2023"

 

The Department for Science, Innovation and Technology Announced An Investment Boost to British AI Supercomputing 

 

A funding boost for two UK supercomputers was also announced by The Department for Science, Innovation and Technology.

These high-performing computers are built with AI capabilities in mind and allow researchers to analyse advanced AI models to test safety features and drive breakthroughs in drug discovery and clean energy.

For context, the new supercomputers will be able to make 200 quadrillion calculations every second. Smartphones deliver only trillions of calculations per second, and a human would have to make a decision every second for 6.3 billion years to match what this computer can calculate in one second.

A new Cambridge supercomputer called “Dawn” will target breakthroughs in fusion energy, healthcare and climate modelling, and an existing supercomputer in Bristol, Isambard-AI, will receive a £225 million investment.

These new supercomputers will be connected to give researchers access to resources with more than 30 times the capacity of the UK’s current largest public AI computing tools. 

 

A New AI Safety Institute was Formed

 

Prime Minister Rishi Sunak also announced a new AI safety group, an evolution of the UK’s Frontier AI Taskforce announced in April 2023.

The AI Safety Institute will be the first state-backed organisation focused on advanced AI safety for the public interest. The idea behind the institute is to “prevent surprise to the UK and humanity from rapid and unexpected advances in AI.” This will include rigorously testing systems before launch, particularly where capabilities are not yet fully understood. 

Ahead of new powerful models expected to be released next year whose capabilities may not be fully understood, its first task will be to quickly put in place the processes and systems to test them before they launch – including open-source models.

The government has also pledged better governance and regulation for AI and to “take action to make sure AI developers are not marking their own homework when it comes to safety.”

Picture: an image showing the word AI repeated on a mustard-coloured background. Image Credit: Unsplash

Article written by Ella Tansley | Published 13 November 2023

Share

---

---

Related Articles

---