convexfunction

6 ms·

1. 121.▲
   by convexfunction 12y ago

   Shameless plug for a (much longer and broader) thing I wrote on sort of the same subject: http://convexfunction.com/blog/2015/01/11/cognitive-behavior...
2. 122.▲
   by convexfunction 12y ago

   Indeed, heuristics and biases have evolutionary origins and served their purpose in the EEA (it really couldn't be any other way...), and indeed, manipulating people into correcting for these heuristics and biases without their knowled
3. 123.▲
   by convexfunction 12y ago

   I didn't see anything about sentience in the article. More generally, sentience has nothing to do with AI safety; whether or not the thing has qualia has very little to do with what the thing may or may not do to our civilization.
4. 124.▲
   by convexfunction 12y ago

   Or maybe that's anthropomorphizing and they won't! Let's just throw the lever when the time comes and find out, and urge people to not spend any time thinking about it too hard until then. This is a good and reasonable and sa
5. 125.▲
   by convexfunction 12y ago

   Andrew, in that case, is definitely right in that long-term AI safety has almost nothing to do with near-term AI implementation. More generally, when I see people assert that these people don't know what they're talking about, it
6. 126.▲
   Catastrophizing at the Scale of Big Data(convexfunction.com)

   2 points by convexfunction 12y ago|0 comments