It did not. In one case it had the number of dislikes correct, but the other cases showed it out by up to 50% - and these were videos that had dislikes registered before youtube disappeared them.
Out of examples, only one was majorly off and these were from a channel with a very limited number of like / dislikes. In other examples the share of likes / dislikes was roughly not too much off.
The video linked is also a small YouTuber with not many views on their videos. Looking at the comparisons it actually gave a pretty close count. Sure it's not exact, but considering it's a small niche YouTuber it seems to be giving a fairly decent approximation. On even larger YouTube channels it's likely going to give you an even better sample of whether people like or dislike the video.
