Why do companies still set algo challenges?!

This annoys me greatly and part of the reason i am hesitant to seek out new challenges.

What annoys me is when i look at these algo challenges, the problem statement is mad for a start, it takes me 10 minutes to figure out what the problem even is lol

Yes i could practice more and spend more time doing katas and the like, but im not convinced its a good indicator anymore. If i know all the short methods built into languages, great, does that mean i am any good at creating real world systems?

Anyone offer any insight to why they still run them and don’t make an assessment resemble what day to day actually looks like?