Can AI lead us to extinction?/ 350 experts raise the alarm

2023-06-14 08:15:00, Kosova & Bota CNA

There are a series of alarms raised by experts in the field of artificial intelligence.

Can artificial intelligence (AI) destroy humanity?

A Brussels-based think tank called PauseAI believes so, and they've been trying to lobby the European Union to stop the spread of more powerful artificial intelligence systems.

In recent months, new tools like ChatGPT have raised fears that AI will lead to millions of people losing their jobs or generating disinformation, especially during elections.

But according to the nearly 300 members of the activist collective PauseAI, these powerful systems can quickly outwit and even manipulate humans.

And if people try to interfere or shut them down, these technologies can resist. Launched in May, PauseAI is the brainchild of software engineer Joep Meindertsma. He took a sabbatical from his current role as director of a technology company to focus solely on PauseAI.

A call for an international summit on AI

In concrete terms, what scares Joep Meindertsma the most is that AI will be able to find zero-day vulnerabilities in cyber affairs.

Zero-day exploits are security vulnerabilities found by hackers in software and systems.

Considering that companies and governments could eventually give AI systems more autonomy and connect them to some vital infrastructure like power grids or even weapons, this could mean that superintelligent AI systems could close them causing chaos.

That's why the founder of the PauseAI group is also asking the EU to lead an international summit on the topic.

More than 350 technology experts sound the alarm

PauseAI is not alone in raising the alarm on this issue. In June, Rishi Sunak, Prime Minister of the United Kingdom became one of the first world leaders to acknowledge the potential "existing" threat of developing a "superintelligent" AI without adequate safeguards.

Recently 350 executives, researchers and engineers working on AI signed an open letter warning that it could one day destroy humanity.

But some argue that AI is still too underdeveloped to pose an existential threat, and that short-term problems like biased and inaccurate responses are the bigger issues right now./ CNA