Why AGI Could Become an Existential Threat to Humanity

Today, we increasingly hear that artificial intelligence (AI) has the potential to change the world. But how seriously should we take this future? I recently watched an interview with AI safety expert Professor Roman Yampolskiy, and his perspective is, to put it mildly, alarming.

AGI and the Existential Threat

According to Yampolskiy, if humanity creates artificial general intelligence (AGI) that surpasses human capabilities in all areas, we may lose control over it. This is no longer just about AI replacing workers or making translators obsolete. It's a far deeper issue.

Why Might Control Become Impossible?

• It can't be "fired"
• It can't be properly punished
• It can't be reliably shut down
• We can't be sure it's acting in our best interests

And this isn't science fiction—it's a real threat that AI safety experts are discussing.

Narrow AI vs. AGI

Currently, we see AI assisting with everyday tasks: writing emails, analyzing data, automating routine processes. People have grown accustomed to the utility of narrow AI tools and may not perceive the risks that the development of AGI could bring. However, it's crucial to understand the difference between these two levels of technology.

Why is Narrow AI Safe?

Narrow AI operates within predefined tasks, lacking the ability to exceed them, and serves as a tool that humans can control. But once we talk about creating a universal intelligence, the situation changes dramatically.

The Future of AGI: Risk or Progress?

If the negative scenarios of AGI come true, we will need to ask ourselves: are we creating a system that we cannot stop? This is a critical distinction we must consider when discussing the future of artificial intelligence.