Does artificial intelligence deserve its doomsday clock?

In 1947, scientists at the University of Chicago (USA) created a concept clock. The midnight hour would be the sign of the end, or the Apocalypse. The Apocalypse. This clock was first published in the Bulletin des savants atomistes in the New Window during the Cold War. It reads 23:53 or “midnight minus 7” when it was made in 1947. It is now 23:58, 58 minutes 30 seconds in 2024.

The hands are moving faster because of current events, like wars, climate change, and nuclear proliferation, while other positive things, like the Fall of the Berlin Wall or the Strategic Arms Reduction Treaty signed between the USSR in 1991, are reversing the pace.

A measure of risk transposed to the subject of artificial intelligence

This AI Safety Clock has been created by IMD Lausanne in Switzerland.

The clock has four sections: Low, Moderate, High and Critical Risk. This assessment is based on the publication of 1,000 specialized websites and 3,470 AI-related news feeds, as well as academic work.

What time are we, according to these academics ?

In 2024, atomists were only 23 hours, 58 minutes and thirty seconds away from the Apocalypse. IMD Lausanne calculates that AI will be here in 23 hours, 31 minutes. It may appear that this is a risk estimate for a distant event.

AI is judged on three criteria: its reasoning and problem-solving abilities, its autonomy, or its ability to work without human interference, and finally, its interaction with real life, such as physical infrastructure, social networks and even weapons.

We must not forget the fact that the main stakeholders in the field of atomic energy were the states via their military personnel. AI is a relatively new technology, with largely private actors. Its implementations are gaining popularity in the public, the commercial and economic spheres. The pendulum will swing in 2025.

Automate with OpenAI and ChatGPT