China, the United States and Europe have all signed, how powerful is this artificial intelligence declaration| Kyoto Brewing Hall

The difference in regulatory purposes and standards greatly reduces the value of the meeting.

▲The world's first AI Security Summit was held in Bletchley Park, UK from the 1st to the 2nd. Photo/Xinhua News Agency

Text | Pottery short room

According to Xinhua News Agency, the world's first artificial intelligence security summit was held in Bletchley Park, England from the 1st to the 2nd, and the summit issued the "Bletchley Declaration" (hereinafter referred to as the "Declaration"). The Declaration argues that the conscious misuse or unintentional control of cutting-edge AI technologies could pose significant risks, particularly in cybersecurity, biotechnology, and the increased spread of disinformation.

This is the first time that the side effects of AI and "the most pressing risks posed by rapidly evolving technologies" have been acknowledged and acknowledged through an internationally consensual approach, acknowledging that "international action is needed" to set a common regulatory approach.

The grand scene of Bletchley Manor

Bletchley Manor, 55 miles north of London, was an early mecca for artificial intelligence. In 1941, it was here that the famous scientist Turing and his code-breaking team successfully deciphered the so-called "undecipherable" Nazi Germany's Engneema cipher, laying the foundation for the victory of the anti-fascist camp in World War II. The choice to hold the world's first summit on AI security here is of great symbolic significance.

According to the organizers, the main goal of the AI Summit is to find some degree of international coordination when agreeing on ethical and responsible principles for the development of AI models, with a focus on so-called "cutting-edge AI" models, which seek to address two key categories of risks in AI through international coordination: misuse and loss of control.

Rapid advances in machine learning, including the emergence of chatbots such as ChatGPT, have prompted governments around the world to consider regulating AI.

The so-called "abuse" risk refers to the risk of bad actors being helped by new AI capabilities. For example, cybercriminals can use AI to develop a new type of malware that security researchers can't detect, or to help countries develop dangerous bioweapons.

The so-called "out of control" risk refers to the fact that artificial intelligence originally created by humans and used to help humans may get out of control, and turn to the detriment of some human groups, or even all mankind.

The participants at the summit came from the European Union and 28 countries around the world with rapid development of artificial intelligence.

In addition to the host British Prime Minister Sunak, the "big shots" include UN Secretary-General António Guterres, European Commission President Ursula von der Leyen, Italian Prime Minister Meloni, US Vice President Harris, and China's Vice Minister of Science and Technology Wu Zhaohui.

In addition, Microsoft President Smith, OpenAI CEO Altman, Tesla CEO Elon Musk, Google DeepMind CEO Hassabis and other business people were present.

Michelle Doniland, the UK's Secretary of State for Science, Innovation and Technology, the organizer of the summit, said it was an achievement to bring together so many key figures in the field of AI. Wu Zhaohui, China's vice minister of science and technology, also said that China would "contribute to an international mechanism on artificial intelligence, expanded participation and a governance framework based on broad consensus to benefit the people and build a community with a shared future for mankind".

The U.S. has also expressed a desire to expand information sharing, research and cooperation, and policy coordination globally. Other participants demonstrated the importance and necessity of strengthening coordinated regulation in various countries in the field of AI from different levels and perspectives.

Notably, the summit was reportedly the brainchild of British Prime Minister Rishi Sunak, who wanted to shape the UK's post-Brexit role, acting as an intermediary between the economies of the US, China and the EU.

The summit is just the first in a series of similar summits. Six months later, it will be held for the second time in South Korea, and the third will be tentatively scheduled for next year (another six months) in France.

▲British Prime Minister Sunak spoke at the summit. Photo/Xinhua News Agency video screenshot

They are all "reasoning" and the reasoning is different

While Sunak tried to seize the "commanding heights" of the international consensus on AI regulation by hosting the summit, calling the summit "a landmark achievement in which the world's most powerful AI powers recognise the urgency of understanding the risks of AI and helping to secure the long-term future of our children and grandchildren", more observers have been calm enough.

The Guardian pointed out that the convening of the summit shows that countries are beginning to realize the dangers of uncontrolled development of artificial intelligence, as well as the need for international coordinated regulation, and begin to agree that countries must "reason" together on this issue. The problem is that governments and industries have very different "truths" to tell.

Recently, U.S. President Joe Biden signed an executive order on artificial intelligence, which proponents claim is "addressing the threat posed by artificial intelligence technology."

However, many industry insiders and experts pointed out that the 63-page executive order "avoids the important and trivializes" and "confuses priorities", which not only pretends to be a powerful force in congressional constraints, but also emphasizes that the government will come forward to establish an "artificial safety research institute" to ensure that the United States is "rightfully leading" in the field of artificial intelligence through government subsidies and talent policy tilt.

In fact, it was not the United States that was the first to adopt a similar executive order, but the United Kingdom, the host of this summit. But the British version of the executive order is also full of "both and all want" contradictions.

While enumerating a series of operational and unpredictable administrative constraints, the UK announced a £225 million investment in an AI supercomputer called Isambard-AI, £100 million for AI security research, and plans to establish an AI security institute. An additional £100 million was invested in the development of AI for healthcare to "compete for the commanding heights of artificial intelligence in Europe".

The European Union, wary of the dangers of AI earlier than other countries, wants to finalize the world's first AI-specific legislation, the AI Act, by the end of 2023 and aim to pass it before the European Parliament elections in June 2024.

The bill is known for being "stricter" – and this is in direct opposition to the UK's stance that the summit's host "can't be choked" and "both regulated and supported". Despite this, the ministers of Germany, Italy and France recently met in Rome to call for a "conducive and innovative approach" to regulate artificial intelligence in Europe, while urging greater investment to meet the challenges of the United States and China in this area.

Representatives of companies and R&D institutions also have mixed opinions, ranging from AI startups that emphasize "not hindering technological progress" to Musk, who openly compares AI risks to a flood of beasts. Their involvement is inevitable, but it is enough to make the already inconsistent "rationale for the implementation of internationally coordinated regulation of artificial intelligence" even more divergent.

More worryingly, US Commerce Secretary Raimondo highlighted the "incredible leadership" of the United States in ensuring AI security during the meeting, while British politicians repeatedly stressed that "an ethical yardstick must be set for AI development".

In this regard, some international media have pointed out that international coordination of a complex and multifaceted technology like artificial intelligence may be difficult. One commenter on Nature's website put it even more sharply – "Moral yardstick? Whose morality counts? ”

▲Data map: The staff (right) is demonstrating a robot based on 5G network to achieve remote control. Photo/Xinhua News Agency

It's better than not at all

International regulation of AI is far from industry and technology consensus. At the same time as the Bletchley summit, many of the two worlds gathered in London for a separate series of events called "AI Fringe", some of which haunted the two venues.

The difference in regulatory purposes and standards also greatly diminished the value of the meeting, which is far from being as important as Sunak has tried to portray. Apart from Sunak himself, there was only one head of state and government, few ministerial-level guests, and even close neighbors such as France and Germany did not send heavyweight officials to attend the meeting.

Despite the disagreement, most participants and observers believe that AI will only develop better if it is put on a standardized track.

Written by Tao Short Room (Columnist)

Editor / Chi Daohua

Proofreading / Wang Xin

China, the United States and Europe have all signed, how powerful is this artificial intelligence declaration| Kyoto Brewing Hall

China, the United States and Europe have all signed, how powerful is this artificial intelligence declaration| Kyoto Brewing Hall

Read on