Recursive AI self-improvement may happen in an "immediate future". Then we enter the phase of "the possibility of us all dying"

Evan Hubinger, a researcher resigning from a US artificial intelligence company, states that AI could kill all people within the next few years. The revelation was made on social media.

This warning came after another researcher, Jacob Coxon, expressed concern about the development of the technology.

In an interview with CNN International, Coxon explains why we should all be worried about the recursive self-improvement of AI, which according to the researchers could happen in an immediate future.