Can't believe Claude Code was launched less than a year ago. It's certainly a new tool in my kit and allowed me to approach and work on new coding problems in a new way.
Great comment and very true in this AI world. In 2030 it will be even easier to make even more realistic images much quicker...
Reminds me of the attacker vs defender dilemma in cybersecurity - attackers just need one attack to succeed while a defender must spend resources considering and defending against all the different possibilities.
Doubtful that they can do this in a way that accurately and effectively helps the user compared to showing the dislike count.
YouTube is plagued by low quality content with clickbait titles, descriptions, and images. Often they outright lie about the content. The recommendation algorithm prioritizes these videos first.
Users can't determine in advance that these videos are poor quality, so they'll be forced to watch those poor quality videos until they find a good one. YouTube wants it to be like this because it increases time on the platform and their advertising revenue.
Along the way, users can dislike these videos but that video still gains views which helps push itself upwards in the recommendation algorithm. Particularly videos with clickbait titles, descriptions, and images tend to amass large numbers of views in short periods of time, which YouTube may recognize as "going viral" and give it an additional push in its recommendations when searching for important keywords.
Furthermore, many users are also watching these videos whilst not logged in or don't care to click dislike, which is another lacking signal to help tune YouTube's ranking algorithm correctly.