3 ms·
This is a good critique. I'm very skeptical of the near term (<100 years) risk of AGI, but I don't really think this article's arguments are valid. Saying we al
by throwawaymath 7y ago
This is a good critique. I'm very skeptical of the near term (<100 years) risk of AGI, but I don't really think this article's arguments are valid. Saying we already have AGI because there exist humans is vacuous and seems to almost deliberately miss the point.
If you want to counter the arguments that AGI will be capable of exponential self-improvement, you need to use an analogue other than humans. Humans categorically lack the capability to exponentially self-improve. Likewise human intelligence is definitionally non-alien, which is not something we can say a priori about any successful AGI we create.