Neutrinos need a large detector volume for efficiency because they interact so rarely. You can’t detect them directly so you need a transparent medium to detect their collision byproducts. Good detector mediums are water and ice, and are underground to minimise background light. There are relatively few places you can do this. Mine caverns and under the sea are the most common, but marine detectors are notoriously hard to build. Francis Halzen proposed using ice. At the Pole, the glacial plateau is 2 miles high and the breakthrough was confirming that the ice is in fact highly transparent if you go deep enough.

Why the pole specifically? You could probably build a second IceCube 100 miles away, but how are you going to get that materials there? Pole has a skiway for large aircraft and infrastructure to house a large number of people. The traverse (SPoT) only became operational near the end of construction - the initial holes were drilled in 2005 and almost everything had to be flown in.

Telescope In The Ice does a great job of explaining the history and science behind the experiment.

The best science comes from areas that have no obvious use. The original insights into nuclear physics, quantum physics and relativity were all pure thought experiments. They led to the world we live in.

I'm personally very happy that we're still funding science that isn't obviously monetised.

The reason for Antarctica is that it's the only place you find cubic kilometres of stable ice that doesn't drift around.

My best guess would be just the remoteness.

1: Ice-mass that isn't contaminated by atomic testing fallout (if the ice has been there since the 40s it's without radioactive waste (faschinating to read about how WW2 wrecks is a prime resource of steel since it's a huge amount of steel without trace amounts of radionucletiods from testing fallout).

2: No current interference (nuclear power, radiowaves,etc) creating possible test uncertainties.

3: I'm sure there are other reasons

The negative naturally is cost, but since the expriment succeeded it will probably be useful for any attempts to send robots or even humanity into space beyond the solar system.

Where else would you go look for a cubic km of ice?

If we're serious about AI taking over jobs, we'll probably need to get more comfortable with any kind of undirected work-for-its-own-sake, not less.