Standard interview format accross industry seems to be

• 4 rounds of algorithm & data structure coding questions with 45 minute te limits
• 1 round of behavioral questions.

I understand algo and coding questions are best way to objectively filter out (less interviewer implicit bias).
But what it entails is candidates have to prepare gours and hours for these questions , even though in industry they are not used that frequently and no one is expecting 45 mins (realistically 30mins as 15 mins are lost in other stuff) to solve anything that requires algorithm in those timelines in actual job role.

I suggest another approach,
where actual real world problems are given to solve and approach to that is used to evaluate,though removing implicit bias is a hard problem to solve.

realistically the second solution will help those who don't have time to prep all the time , either because they are not very actively looking or have family responsibility that takes away time.

With behavior rounds i have not seen any obvious disadvantages to anyone except there is some chance of implicit bias (this can be mitigated based on how companies record their interview responseands and proof them)

From buisness side of things why current model might not be efficient: Company may get candidates that can solve leetcode type problems but may loose out on high intution thinkers who think out of box.
I am not saying they are mutually exclusive but statistically intutive and out of box type personality has hard time conforming to rules.

Disclaimer: I have been facing these issues , so this opinion may be biased in my favor. I tried to be unbiased as possible.