YouTube has introduced several new features recently; many of them are based on Artificial Intelligence (AI). A YouTuber has now presented a feature that should actually support content creators on the platform, but fails in its implementation.
Who is the YouTuber? It is the American YouTuber Drew Gooden. He started on the short video platform Vine and uploaded commentary videos to YouTube after the app was shut down.
In these videos, he is mostly ironic, but this is now humorously backfiring on him with a YouTube AI feature.
In the German-speaking YouTube space, HandOfBlood is an important part. We present him to you in the video:
AI feature should make it easier for YouTubers to respond to comments
What kind of feature is it? In addition to AI-generated summaries of videos, there is also a feature that gives YouTubers the ability to provide AI-generated responses to comments. Initially, this feature was very generic: It suggested phrases like “Thank you”, “Glad you liked it” or “Sorry”.
The feature was presumably intended to help YouTubers respond to comments under their videos faster without having to write much.
However, the feature was then updated to be even more accurate. The generated responses are now tailored to the individual style of the respective YouTuber.
The suggested responses are, however, rather disappointing and amusing for Gooden than actually useful. Drew Gooden noted this in one of his YouTube videos.
AI suggests many inappropriate responses
What kind of responses does YouTube generate? For a comment, the function, for example, comes up with a nasty response and suggests responding to a slightly longer comment with “That is a very long comment” or “I don’t care”.
When it comes to comments that have a serious tone, YouTube seems unable to grasp this. For example, a user talks about his bad experiences with gambling, and YouTube suggests commenting with “That’s great!”.
Drew Gooden believes that what makes this feature so useless is the fact that most responses consist of a variation of “I don’t know”:
- „I don’t know what that means”
- „I don’t know how to use that”
- „I don’t know what you are talking about”
Drew Gooden says in his video: “Even if that were my actual reaction, why do I need your [the AI feature’s] help for that? Am I so dumb that I can’t even think of ‘I don’t know’ on my own!?”
However, the feature not only portrays Drew Gooden as dumb but also as vain. Some suggestions for compliments in the comments are the following:
- User: „Top-tier video“ – Suggestion: „I know, right?”
- User: „Good video!“ – Suggestion: „Yes, it was so funny.”
- User: „Such an important video“ – Suggestion: „Yes, I know it’s a big deal”
AI apparently does not recognize irony and sarcasm
What is the reason for that? YouTube draws its information to create responses in the sense of the YouTuber from the transcript of the video. The problem: If a creator is sarcastic or ironic in their statements, the function does not recognize this.
This leads in the AI-generated responses to Drew Gooden appearing as a dumb narcissist. He himself takes the whole thing in stride and says: “It’s really funny that they want me to be a huge narcissist and a complete idiot. They’re really not doing me any favors by imposing that on me.”
This AI feature from YouTube is quite amusing, but other functions in different areas can quickly become dangerous. MeinMMO author Benedikt Schlotmann addresses the dangers of AI in this article: Researchers warn of 700 options for AI failures: 3 particularly dangerous ones have already occurred
