lhe Paris Peace Forum (November 11-12), on the sidelines of which a meeting was organized on Sunday, November 10 between President Macron and an advisory group on artificial intelligence (AI), is taking place in the shadow of continuing wars in Europe and the Middle East.
Russian and Ukrainian Telegram channels are filled with videos of drones, both military models and DIY consumer models, tracking fighters trying to escape their high-pitched drone before being incinerated before the eyes of thousands of viewers. Several news reports also revealed that Israel used an artificial intelligence system to select targets in Gaza.
We are at a major turning point in warfare, about to move from a war waged by humans assisted by AI to a war waged by humans assisted by AI. Such a change, in which AI would make the final decision to launch lethal attacks without human intervention, would mark a new dark age for our species.
Removing humans from decision-making could lead to unprecedented levels of efficiency in war crimes, eliminating the possibility of junior officers questioning orders to stop massacres of unarmed civilians or combatants who have surrendered in accordance with international law.
Massive losses
Artificial intelligence systems can evolve at an incredible speed and a programming error could cause massive losses even before human supervisors realize it, or lead low-level conflicts to an escalation that was not the intention of the decision makers. political decisions.
These risks weigh not only on the ground, but also on cyber weapons. HE malware [« logiciels malveillants »] States have been using autonomous and automated systems for a relatively long time. The United States’ development of the Stuxnet virus, which successfully crippled Iran’s nuclear program, likely began in 2005.
But these tools often required detailed knowledge of target systems through human intelligence (as with Stuxnet) or had to be widely distributed to be effective (such as NotPetya and WannaCry, distributed by Russia and North Korea).
When it comes to cyberattacks, AI has so far primarily benefited defenders, like companies like mine. [SentinelOne] Use it to detect and stop attacks in real time.
You have 60.65% of this article left to read. The rest is reserved for subscribers.
