I understand some of the inherent dangers involved with AGI and advanced machine learning. While I can see some of the more low-level risks associated with AI coming to fruition (deep-fakes, biased algorithms, etc.), some of the more existential dangers seem far-fetched and lack empirical examples. As someone outside of AI development circles and without a background in comp-sci, I would like to know how realistic those working in the field of AI find these fears. Are there any convincing hypotheticals about the risks of a superintelligence?
I've read some Stuart Russel and have recently been watching Robert Miles videos on the topic. Both of these observers seem worried. Is this opinion widely held or is it seen as a bit extreme among developers?
Provably Beneficial Artificial Intelligence by Stuart Russell