I’m currently studying a statistics PhD while researching AI safety, after a bioinformatics msc and medical undergrad. I agree with some parts of this, but would contest others.
I agree that:
What you do within a major can matter more than which major you choose
It’s easier to move from math and physics to CS.
But it’s still easier to move from CS to CS, than from physics or pure math. And CS is where a decent majority of AI safety work is done. The second-most prevalent subject is statistics, due to its containing statistical learning (aka machine learning) and causal inference, although these are areas of research that are equally performed in a CS department. So if impact was the only concern, starting with CS would still be my advice, followed by statistics.
I’m currently studying a statistics PhD while researching AI safety, after a bioinformatics msc and medical undergrad. I agree with some parts of this, but would contest others.
I agree that:
What you do within a major can matter more than which major you choose
It’s easier to move from math and physics to CS.
But it’s still easier to move from CS to CS, than from physics or pure math. And CS is where a decent majority of AI safety work is done. The second-most prevalent subject is statistics, due to its containing statistical learning (aka machine learning) and causal inference, although these are areas of research that are equally performed in a CS department. So if impact was the only concern, starting with CS would still be my advice, followed by statistics.