The China Mail - India hosts AI summit as safety concerns grow

While frenzied demand for generative AI has turbocharged profits and share prices for many technology companies, anxiety is growing over the risks that it poses to society and the environment.

Prime Minister Narendra Modi will on Monday afternoon inaugurate the five-day AI Impact Summit, which aims to declare a "shared roadmap for global AI governance and collaboration".

It is the fourth annual gathering addressing the problems and opportunities posed by AI, after previous international meetings in Paris, Seoul and Britain's wartime code-breaking hub Bletchley.

Touted as the biggest edition yet, the Indian government is expecting 250,000 visitors from across the sector, including 20 national leaders and 45 ministerial-level delegations.

Also in attendance will be tech CEOs including Sam Altman of OpenAI and Google's Sundar Pichai, although unforeseen circumstances have reportedly led Jensen Huang, head of US chip titan Nvidia, to cancel his planned appearance.

Modi will seek to "strengthen global partnerships and define India's leadership in the AI decade ahead" in talks with the likes of France's Emmanuel Macron and Brazil's Luiz Inacio Lula da Silva, organisers say.

But whether they will take meaningful steps to hold AI giants accountable is in doubt, Amba Kak, co-executive director of the AI Now Institute, told AFP.

Industry commitments made at previous events "have largely been narrow 'self regulatory' frameworks that position AI companies to continue to grade their own homework", said Kak, a former AI advisor to the US Federal Trade Commission who is taking part in the summit.

- AI safety -

The Bletchley gathering, held in 2023 -- a year after ChatGPT stunned the world -- was called the AI Safety Summit.

The meetings' names have changed as they have grown in size and scope, and at last year's AI Action Summit in Paris, dozens of nations signed a statement calling for efforts to flank AI tech with regulation to make it "open" and "ethical".

But the United States did not sign, with Vice President JD Vance warning that "excessive regulation... could kill a transformative sector just as it's taking off".

The Delhi summit has the loose themes of "people, progress, planet" -- dubbed three "sutras".

Even so, AI safety remains a priority, including the dangers of misinformation such as deepfakes.

Last month saw a global backlash over Elon Musk's Grok AI tool because it allowed users to produce sexualised pictures of real people, including children, using simple text prompts.

"Child safety and digital harms are also moving up the agenda, particularly as generative AI lowers the barrier to harmful content," AI Asia Pacific Institute director Kelly Forbes told AFP.

"There is real scope for change" although it might not happen fast enough, said Forbes, whose organisation is researching how Australia and other countries are requiring platforms to confront the issue.

- AI for 'the many' -

Organisers highlight this year's AI summit as the first to be hosted by a developing country.

"The summit will shape a shared vision for AI that truly serves the many, not just the few," India's IT ministry has said.

Last year India leapt to third place -- overtaking South Korea and Japan -- in an annual global ranking of AI competitiveness calculated by Stanford University researchers.

But despite plans for large-scale infrastructure and grand ambitions for innovation, experts say the country still has a long way to go before it can rival the United States and China.

Neither Donald Trump nor Xi Jinping will attend the summit, but both countries are sending high-level tech policy officials.

Seth Hays, author of the Asia AI Policy Monitor newsletter, said talk at the summit would likely centre around "ensuring that governments put up some guardrails, but don't throttle AI development".

"There may be some announcements for more state investment in AI, but it may not move the needle much -- as India needs partnerships to integrate on the international scene for AI," Hays told AFP.

H.Ng--ThChM