Student AI activists at Encode Justice release 22 goals for 2030 ahead of global summit in Seoul

Sneha Revanur is a 19-year-old incoming junior at Stanford University, but she is already making waves in Washington as an activist on the ethical use of artificial intelligence (AI). Her organisation Encode Justice has put out a new list of goals ahead of this week’s AI Seoul Summit.

Encode Justice last week released its AI 2030 plan, a list of five broad calls to action and 22 specific goals that the students behind the group hope to see addressed in concrete ways by 2030. The goals include issues that have been part of the global conversation around AI but have so far seen little progress on international agreement. These include disclosures for when AI is used in political advertising or when users interact with the technology, ownership over personal data, and preventing the use of automated weapons of mass destruction.

While founded in the US, Encode Justice has participation from students in 30 countries, according to its website. The Generation Z activists have also found some high-profile academics and officials from around the world to join as signatories to the AI 2030 document, including former Taiwan digital minister Audrey Tang.

“Obviously we have a pretty clear case to make here [given that] we’re the generation that’s actually going to inherit the impacts of these technologies,” Revanur said. “So world leaders have a particular duty to us to ensure that … they’re actually keeping [this] in mind as they pursue governance going forward.”

Encode Justice founder Sneha Revanur pictured outside the US Capitol building in Washington. At 19-years-old, Revanur has already been addressing policymakers, and Encode Justice was one of several organisations consulted for the US “Blueprint for an AI Bill of Rights”. Photo: Instagram / @sneha.revanur

Tang – whose term ended with that of Taiwan President Tsai Ing-wen on Monday – said she was drawn to lending her name to the document because of Encode Justice’s “comprehensive approach to ethical AI development, focusing on inclusivity and safety”.

“Guiding AI development from a ‘race to power’ towards a ‘race to safety’ is our shared mission, and I hope this call can raise awareness around it,” she said.

Neil Zhu Xiaohu, the founder of the Centre for Safe AGI (artificial general intelligence) and one of two AI 2030 signatories from mainland China, said he came across the goals when friends shared them on social media.

“Young people should be heard, which is important for the future of humanity,” he said. “I hope people can turn to a safety-first approach in the next few years so that we can navigate the AI era.”

Countries around the world are currently racing to find ways to grapple with concerns raised by AI around issues such as ethics, safety and privacy. Entrepreneurs such as Tesla CEO Elon Musk and OpenAI CEO Sam Altman have warned of potential existential threats from AI, as have researchers like groundbreaking AI computer scientist Yoshua Bengio, another AI 2030 signatory.

This has led to the latest AI summit in Seoul taking place on Tuesday and Wednesday, after an AI Safety Summit in the UK last November.

However, the leading markets for AI, the US and China, are constantly at odds and have different priorities when it comes to regulating the technology. The US has a blueprint for an AI “Bill of Rights” and some rules for companies selling to the government, but not comprehensive regulations. China, on the other hand, requires that large language models (LLMs) get approval for public release and has laws protecting user data.

But China’s approach to the development of LLMs – the technology behind ChatGPT and other generative AI services – “could have chilling effects for free speech”, Revanur said.

“We’re excited about the fact that clearly policymakers in China are taking this seriously. We hope to use that to our advantage to build international consensus,” she said. “But … there are some tension points when it comes to the values and priorities that we are guided by and the values and priorities that might be guiding some of these frameworks.”

Encode Justice seeks to enhance the opportunities available from AI rather than regulate the technology away. At the same time, some of its goals for 2030 have a clearer path to realisation than others.

An international treaty on the prevention of fully autonomous offensive weapon systems, for example, could be modelled after something like the Convention on Certain Conventional Weapons, according to Encode Justice. Others have said AI needs something akin to the Treaty on the Non-Proliferation of Nuclear Weapons.

“I think reducing potentially catastrophic risks is an area that both [the US and China] have been quite open in dialogue about,” said Luke Drago, a member of the Encode Justice board based in the UK. “This is an area that we think is probably primed for international cooperation.”

Sam Altman, CEO of ChatGPT creator OpenAI, participates in a panel discussion during the World Economic Forum in Davos on January 18, 2024. Altman has previously said that AI poses an existential risk to humanity. Photo: AP

Brian Wong, a political philosopher at the University of Hong Kong who also works on AI governance, said at least some kind of agreement is needed “to not deploy AI in lethal, or potentially lethal, weaponry in high-intensity, low-confidence, high-uncertainty contexts”.

“Beijing and Washington have been ratcheting up the frequency and depth of … conversations on AI alignment,” he added. “I find that encouraging – though more really must and can be done.”

Another major concern of the document is the impact AI will have on work and income inequality. It asks that particular attention be paid to the Global South, which could be hardest hit, and calls for governments to set up a global retraining fund.

The sweeping scale of the AI 2030 document could make achieving everything a daunting task. Revanur also pointed out that Encode Justice members have encountered “age bias” affecting the “perception of our work”.

Wong, who knows some members of Encode Justice, said the AI 2030 goals are not a case of overreach from a student movement. “There are some fairly well-qualified, for their ages, and thoughtful voices in the leadership teams of the movement,” he said.

Still, the AI 2030 goals are ultimately an act of optimism.

“We’re very optimistic about the potential [of achieving these goals],” Drago said, “especially because AI applications can uplift emerging economies through innovations in healthcare, agriculture, education, it can bring down costs in major innovations for emerging economies as well.”

Revanur summed up the driving force behind Encode Justice succinctly: “We’re young; it’d be a crime not to be optimistic.”



Read More

Leave a Reply