With the rise of google re-Captcha that is known to be the most effective one because of the use of real images that are computational hard to get distinguished from a machine. In other words a human is easily able to distinguish a truck from a duck whilst for a comptuter is rather hard or computational expensive to do that.
But I still wonder how old good text based captcha implementations like the one that gitea uses (look image bellow) are good enough to prove that the form has been submitted by a human? Or image recognition is the most efficient nowdays that these type of caphca tests has been rendered as junk?