Hacker News new | past | comments | ask | show | jobs | submit login

You're not the only one who finds it scary, as there are massively popular books on the topic..

https://www.amazon.com/Superintelligence-Dangers-Strategies-...




I found it difficult to make it through the first couple of chapters.

Having just read The Gene, his analysis of the artificial selection option for super intelligence came across as very wrong - underestimating the complexity of polygenic traits and making (likely inaccurate) assumptions about their heredity. This type of thing has a dangerous history.

The idea that a super intelligence could emerge from the internet unexplained also seemed pretty weak, but the first issue was so bad I found it hard to take anything else seriously (didn't trust the author's analysis).

There are interesting issues with artificial consciousness, but I think they're in some ways similar to the issues with biological consciousness - the testing data the neural net is exposed to and it's underlying model can lead to minds that wouldn't be considered intelligent (and dangerous outcomes as a result).


I would suggest giving it another go. There are few authors who have given this issue as much thought as Bostrom, and if the conclusions he draws are false at the very least that opens up the door for further conversation about the subject.




Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: