A half-joking prediction:
at some point we'll solve all arbitrarily hard milestones for AIs and will still find ourselves 'nowhere near having real general intelligence'.
At that point we might start questioning our assumptions about intelligence.
it'll be the singularity, superintelligence will be optimizing human affairs and launching probes to other stars as it disassembles the solar system for material to build a Dyson swarm, and people will still claim it's not "really" intelligent.
- shared context (shared biological & cultural heritage)
- recognition & willingness to ascribe intelligence
Basically trying to imply that the general AI problem is of similar nature to the 10x programmer problem - communication & recognition, or lack of thereof.
A 10x programmer makes a hard problem look easy. A general AI makes people around it feel very smart & productive.
Which is also why I prefaced it with, The terrain is not the map - we humans know surprisingly little about our intelligence.
[edit]
Case in point, an older discussion about intelligence in animals: https://news.ycombinator.com/item?id=21772648