Source: https://www.youtube.com/watch?v=MWHN6ojlVXI
***
The main reason why global tech leaders are afraid of AI disaster is not really AI going rogue and attempting to destroy humanity. In fact, we're afraid that **AI might give us exactly what we ask, for instance taking steps towards a certain outcome that would be detrimental to humanity.**
**It's called specification gaming.**
On the other hand, positive impact we expect from AI at a large scale is to allow humanity to make some **frog leaps towards discoveries** that would normally be centuries ahead of us.