Am a computer scientist, working on AI alignment theory.
I’m probably one of the people where I work who is more sympathetic to MIRI-style ways of thinking about alignment.
Leaned towards a type of thinking that I labelled “algebraic” as a math undergrad.
My best course in undergrad was intro to analysis, but it was taught by a PDEs guy. Our department only had one real analyst, and was predominantly composed of algebra people.
My favourite take on linear algebra involves a ‘geometric’ approach, e.g. thinking of linear operators, not matrices, and taking this sort of view of the singular value decomposition.
I wish that everybody would always denote vectors with bra-ket notation.
My primary academic contribution to CS was to take a bunch of proofs about one family of probability distributions, and see if they worked on a different family of probability distributions (if this doesn’t sound CS-y.… uh, I’m a fake CS boy).
I really like Haskell, and deviations from it really bother me. In particular, the bits I like are the fact that it’s functional and strictly typed.
I mostly use Python because it’s easier.
Object-oriented programming seems weird and creepy to me.
I use emacs, and have briefly tried vi-type things but they never stuck.
When evaluating arguments, I tend to ask questions like “is this argument symmetric in the appropriate variables”, “if you take this variable to 0 or infinity, does the argument still work”, or “does this type check”. I could translate this into terms that make more sense for verbal/non-mathematical arguments, but honestly this is how I think of it.
When eating corn on the cob, I think I do it in spirals.
I only eat corn on the cob at my family home where I grew up, which is a different part of my life than the part that contains everything else on this list.
Looking over the post, I guess that I’m basically an algebraist except for the way I eat corn?
Am a computer scientist, working on AI alignment theory.
I’m probably one of the people where I work who is more sympathetic to MIRI-style ways of thinking about alignment.
Leaned towards a type of thinking that I labelled “algebraic” as a math undergrad.
My best course in undergrad was intro to analysis, but it was taught by a PDEs guy. Our department only had one real analyst, and was predominantly composed of algebra people.
My favourite take on linear algebra involves a ‘geometric’ approach, e.g. thinking of linear operators, not matrices, and taking this sort of view of the singular value decomposition.
I wish that everybody would always denote vectors with bra-ket notation.
My primary academic contribution to CS was to take a bunch of proofs about one family of probability distributions, and see if they worked on a different family of probability distributions (if this doesn’t sound CS-y.… uh, I’m a fake CS boy).
I really like Haskell, and deviations from it really bother me. In particular, the bits I like are the fact that it’s functional and strictly typed.
I mostly use Python because it’s easier.
Object-oriented programming seems weird and creepy to me.
I use emacs, and have briefly tried vi-type things but they never stuck.
When evaluating arguments, I tend to ask questions like “is this argument symmetric in the appropriate variables”, “if you take this variable to 0 or infinity, does the argument still work”, or “does this type check”. I could translate this into terms that make more sense for verbal/non-mathematical arguments, but honestly this is how I think of it.
When eating corn on the cob, I think I do it in spirals.
I only eat corn on the cob at my family home where I grew up, which is a different part of my life than the part that contains everything else on this list.
Looking over the post, I guess that I’m basically an algebraist except for the way I eat corn?