No consensus among experts on AI risks, trajectory ‘remarkably uncertain’: report

OTTAWA – A major international report on the safety of artificial intelligence says experts can’t agree on the risks the technology poses – and it’s unclear whether AI will help or harm us.

The report, chaired by Canada’s Yoshua Bengio, concludes that the “future trajectory of general-purpose AI is remarkably uncertain.”

It says that a “wide range of pathways” are possible “even in the near future, with both very positive and very negative outcomes.”

The report was prepared at last year’s AI Safety Summit hosted by Britain, the first global meeting on artificial intelligence.

Britain tapped Bengio, a “godfather” of AI and scientific director of Mila, the Quebec AI Institute, to chair the report. It was released ahead of a new global summit on AI, to be held next week in Seoul, South Korea.

“We know that advanced AI is developing very quickly and that there is significant uncertainty about how these advanced AI systems may impact the way we live and work in the future,” Bengio wrote in the report.

The British government said in a press release on Friday that the report is the “first ever independent, international scientific report” on the safety of AI, and that it would “play a substantial role” in informing discussions in South Korea next week .

A group of 75 experts contributed to the report, including a panel nominated by 30 countries, the European Union and the United Nations. The report published on Friday is an interim report and the final version is expected by the end of the year.

It focuses on general-purpose AI systems such as OpenAI’s ChatGPT, which can generate text, images and videos based on cues.

The report says experts “continue to disagree on several questions, small and large, around general-purpose AI capabilities, risks, and risk mitigations.”

One of the points of discussion is the likelihood of “risks such as large-scale labor market impacts, AI-based hacking or biological attacks, and society losing control over general purpose AI.”

The report outlines a number of risks, including the damage AI can cause through fake content, disinformation and fraud, as well as cyber-attacks. It also points out the risks that bias can cause in AI, especially in “high-stakes domains such as healthcare, job recruiting and financial lending.”

One possible scenario is that humans lose control of artificial intelligence and no longer have control over the technology, even though it can cause harm.

According to the report, there is consensus that current general-purpose technology does not pose that risk, but some experts believe that ongoing work to develop autonomous AI, which can “act, plan and pursue goals,” could lead to a could lead to such a result.

“Experts disagree on how plausible scenarios of loss of control are, when they might occur, and how difficult it would be to mitigate them,” the report said.

This report by The Canadian Press was first published May 18, 2024.

Anja Karadeglija, The Canadian Press