Arabic version: صحفيو الغارديان يجيبون عن أسئلة مخاطر الذكاء الاصطناعي وتنظيمه

According to The Guardian, technology reporters Aisha Down, Blake Montgomery and Dan Milmo answered reader questions on artificial intelligence after a week of warnings about its potential to destroy the world as people know it.

Montgomery said individuals have limited options when faced with declarations of an AI apocalypse, though they can stop using AI and vote for politicians who support regulation. He said he does not put much stock in an AI-made doomsday, arguing that a genuine belief in it would require drastic life changes today.

On regulation, Montgomery said restricting access to computing power could hinder broader AI development. In the absence of legislation covering every kind of AI development, he said lawmakers may instead target specific areas, including housing allotment and children’s access to chatbots. He said cooperation from China would be necessary for an intentional slowdown to succeed, and that AI would be high on the agenda when Donald Trump and Xi Jinping meet in Washington next week.

Down said there is an element of a “black box” in the way systems such as ChatGPT or Claude respond to particular prompts, even though the broad processes by which large language models are trained and built are understood. She also described the finances of the AI buildout as shaky, pointing to hundreds of billions of dollars going into infrastructure projects, GPU-backed debt and questions over whether expensive Nvidia GPUs are being put into datacentres.

Montgomery said critics argue that focus on a potential AI apocalypse can distract from current harms, including algorithmic discrimination in housing allotment. Max Tegmark of the Future of Life Institute rejected the idea that existential and immediate risks cannot be discussed together, comparing it to refusing to discuss climate change because a forest fire is already burning.