For people's consideration while discussing the dice in this game, I'll return this test that was discussed some time ago, which I cannot now find the thread for any more...https://docs.google.com/spreadsheets/d/1tdyBoQNS_vwEGZGBgFRQex7b-Ma8S6P7zvEMK5wh9n4/edit?usp=sharing
This test was taken shortly before the introduction of the loaded dice 'fix', and illustrates that the RNG Larian is using is not a very good one, in light of being the basis for a game based around the use of a random rolling mechanic for the majority of its important systems.
Rather than fix this problem ,they implemented a hack shortcut to 'load' the dice in the players favour whenever their poorly functioning RNG moves into one of its quite predictable downswings. This doesn't fix the problem and it comes at the cost of the game back-handedly calling the player a cheater, and demeaning them for choosing to have it turned on.
(Edit: to explain the document, the left hand columns are a series 200 consecutive rolls taken under as near to stable conditions as could be achieved, in a series of different roll-based video games. The upper chart on the right shows the various averages and means of each game's results, while the graphs then show a plotting of each game's consecutive rolls; BG3 is at the top, and shows a substantial flaw compared to all of the others.)