BREAKING: Scientists say whatever safeties we create may be useless in preventing harmful AI from taking over the world.
In fact, there is no current evidence that AI can be controlled safely, according to an extensive review, and without proof that AI can be controlled, it should not be developed, AI safety expert Dr Roman V. Yampolskiy warns.
When connected to the internet, AI may have access to all of humanity's data, allowing it to replace all existing programs and take control of all online machines worldwide.
And in a recent study, researchers developed a theoretical containment algorithm, which ensures that a super-intelligent AI won’t harm people under any circumstances. According to the study, an algorithm that would command AI not to destroy the world could destroy its own operation if it detects a potential threat.
However, it remains unclear whether the algorithm has successfully prevented a catastrophic event or is still analyzing the threat, making it difficult to know for certain if the machine has successfully prevented a catastrophe.
The study shows that traditional ideas for controlling super-intelligent AI have limitations, and new strategies are necessary, starting with disconnecting AI from the internet.
Learn more:
https://www.eurekalert.org/news-releases/1032090
.png)