Sometimes I will start reading a paper and go down a rabbit hole following lots of different papers. There is so many papers to read and so many directions to go towards and sometimes I end up getting overwhelmed and frustrated because there is too much information and it all seems too hard.
So sometimes I do have to re-remind myself to take it easy with AGI research.
Its highly unlikely to be solved anytime soon, I mean this is one of the hardest problems since human language appeared, but I will continue to study it as hard as I can. I am working absolutely hard, but I am at ease and if it doesn’t work out, no big deal. Im having fun learning and satisfying my curiosity about AGI. This is a marathon, not a sprint.
I am looking for the right opportunity to strike if and when I see the right technologies appear. Some understanding of how to implement:
- systemic generalization and compositionality [article]
- Sensorimotor representation
- A feasible way to represent grounded/embodied learning in computers
- Amodal generalized representations [what is the test that proves this works]
- Efficient model representations
So just be chill, have fun, keep writing in my zettelkasten, and learn 🙂