Published: Monday, May 20, 2020
SEOUL, South Korea – This week, South Korea will host a mini summit on artificial intelligence risks and regulations, following a similar event held in Britain last summer that brought together a wide range of experts, researchers, and government officials.
The Seoul meeting aims to continue the work begun at the U.K. gathering on reining in threats from cutting-edge artificial intelligence systems.
What you need to Know about the AI Seoul Summit, and AI Safety Issues
What international efforts have been made on AI safety?
The Seoul Summit is part of a global effort to create safeguards for rapidly evolving technology. This technology promises to transform many aspects in society. However, it has also raised concern about new risks to everyday life. For example, algorithmic bias can skew search results.
Researchers, government leaders and tech executives, some of whom held opposing views about AI, met in closed-door discussions at the U.K. summit in November, which was held in a former secret codebreaking base from World War II in Bletchley, north of London. Tesla CEO Elon Mots and OpenAI CEO Sam Altman mixed with politicians such as British Prime Minister Rishi Sunderak.
The Bletchley Declaration was signed by delegates from over two dozen countries, including the U.S.A. and China. They agreed to work together in order to limit the “catastrophic risks” posed by the rapid advances of artificial intelligence.
The U.N. General Assembly passed its first resolution on Artificial Intelligence in March. This was a major step forward for an international initiative to ensure that this powerful technology is beneficial to all nations and respects human rights, while also being “safe, secured and trustworthy”.
In Geneva, earlier this month the U.S. held its first high-level discussions on artificial intelligence to discuss the challenges of the rapidly evolving technology and to set standards for managing it. U.S. officials expressed concerns over China’s “misuse” of AI, while Chinese representatives criticised the U.S. for “restrictions” and “pressure” placed on artificial intelligence.
WHAT IS THE SUMMITE IN SEOUL ABOUT?
The South Korean and British governments will co-host the May 21-22 meeting.
On the first day, Tuesday, South Korean president Yoon Suk Yeol will be joined by Sunak to meet with leaders virtually. Few global industry leaders were invited to give an update on their progress in fulfilling their commitments at the Bletchley Summit to ensure the safety and security of AI models.
The second day will see digital ministers gather in person for a meeting hosted by South Korean Science Secretary Lee Jong-ho, and Britain’s Technology secretary Michelle Donelan. Participants will exchange best practices and concrete plans of action. Participants will also share ideas about how to protect society against the potential negative impacts of AI, such as on energy usage, workers, and the proliferation mis- and deinformation, according organizers.
The mini-virtual summit is a temporary meeting that will serve as a bridge until the full-fledged edition in person, which France has promised to hold.
Representatives from the United States, China and Germany will be present at this meeting, as well as OpenAI, ChatGPT, Microsoft, Google and Anthropic.
What progress has AI Safety Efforts made?
The agreement reached at the U.K. summit was lacking in details and did not propose a method to regulate AI development.
The United States and China attended the last summit. When we examine the principles that were announced following the meeting, some of them are similar to those announced at previous U.N. or OECD conferences,” said Lee Seong Yeob, a Professor at Korea University’s Graduate School of Management of Technology. “There was no new information.”
He said that it’s important to have a global summit to discuss AI safety, but that all participants will find it “considerably hard” to come to an agreement, as each country has its own interests and levels of AI technology and industry.
Meta, OpenAI, and Google are launching the latest AI models at this gathering.
The original AI Safety Summit’s purpose was to find solutions for the so-called “existential risks” posed by powerful AI systems such as ChatGPT.
Yoshua Bengio was selected to head an expert panel at the U.K. conference, which will draft a report about the current state of AI safety. The interim report, released Friday, aims to provide a basis for discussions in Seoul. It identifies a number of risks that general purpose AI poses, such as its use maliciously to increase “scale and sophistication” when it comes to frauds and scams.
The report stated that AI systems with malfunctions could cause bias to spread in fields such as healthcare, recruitment for jobs and financial lending. In addition, the technology’s ability to automate many tasks poses systemic risk to the labor market.
South Korea wants to take advantage of the Seoul Summit to formulate global AI governance and standards. Some critics claim that South Korea’s AI infrastructure is not advanced enough for it to be a leader in governance issues.
__
Chan reported from London.
Source: ABC News