I did not write this comment.

After a conversation with some friends, I figured I'd make a quick video about YouTube's new AI-generated comments.

The 'Capital Cities' Worldbuilding video will be up in a week or so.

Contents:
00:00 Intro
00:30 YouTube Creator Studio Rundown
1:16 Community Tab Curiosities
1:58 AI-Suggested Comments and Concerns
4:30 Too Close
7:11 How did it know?
10:23 Do you care?
12:27 Conclusion
12:55 Outro

[Russell_DM]

To those brought here by the algorithm:

As I mentioned at the start of the video, this isn't really a "current events" channel. This is a channel focused on writing, worldbuilding, and TTRPGs. On that note, I've seen a lot of doom-and-gloom in the comment section, and I would feel guilty if I didn't also attempt to use this video's success to put some positivity back into the world.

Write. Write every day, if you can. It doesn't matter if it's a journal, a fantasy saga, or a fanfic. Just get out there and write. It's fun. It's good for you. It's good for the world. Letting a work of art spring from your mind into existence is one of the most beautiful things a human can do—and that's never going to change.

And, if one or two of you ever get the itch to see what this whole "Worldbuilding" thing is about, consider checking out the rest of the channel. If you're more interested in the "D&D" side of things, I'll direct you to the video that got me started, without which this channel wouldn't exist:

[imjustsomepersonontheinternet]

At the risk of feeding this feature with my comment: I absolutely hate this. I'm a software engineer. AI is fascinating to me. I also fucking hate it. I wouldn't want any AI to generate anything personally identifiable to me ever without my explicit consent.

[TheGuyWhoIsSitting]

YouTube: does this

Also YouTube: “nothing we can do about bot accounts and scams”

[Creative_YT]

Ai comments… defeat the entire point of comments

[TheDanibits]

Every day we come closer and closer to the Dead Internet Theory. AI generated videos, written by AI, being commented on by AI "viewers".

[atreidhd]

Bare minimum, it should be opt-in and all comments written with it should have an identifier next to their timestamps. The reason neither of those happened is because Google knows that the vast majority of people would not consent to it, including the audience.

[robotwizard]

Software engineer here, in respect to what was said at 6:10, they absolutely did plug your stuff into an LLM. LLMs (which all so called "ai" are as of now) can be pretty advanced and Youtube has the infrastructure and data to train it on your specific voice and opinions, they could plug this into an LLM trivially, it was just luck that this specific generated line was nuanced and appropriate. A great number of their suggestions will not be (as seen by the other generated comment). Please do NOT use this feature, even if it makes a good comment, it lets them use human input to inform what "good generated comments" look like which will allow them to farm more data and encourage more investment in LLMs which are all things no one really wants.

[timeittells]

oh dude i saw the title and thought it was advanced scammers. Great video interesting to know

[slouch]

maybe i'm a bit cringe but it means a lot to me when a video creator responds to one of my comments. i like the idea of being able to communicate with someone who made something i enjoyed, and to feel like they cared enough about whatever i happen to have commented to reply. it always improves the impression i have of a creator when they reply to my comment, and makes me much more interested in other things they have or will post.

reply-by-AI feels like a way to trick me. now i am not sure if the reply means they actually care about what i wrote, or if they scrolled down a list pressing a button to autorespond to a bunch of comments. that ambiguity means i basically can't trust any comment i see on YouTube to be authentic anymore. that makes me sad.

[Chimmaney]

Rock-solid agree with everything you say in this video. That YouTube will automatically use your data to create something "like you", without permission or warning, is a disturbing precedent to set. The de-personifying of online content is something that 100% needs to be discussed and controlled before companies can go too far with it.

[bobito]

thanks for whistleblowing this to casual users

[remor]

What makes this even worse is that it amplifies the problem of the parasocial aspect of social media platforms.
Before, it was a question of whether or not you were having a consistent back and forth with another person that could tell you if you could genuinely call yourselves acquainted to at least a passing extent.
Now, as a commenter, you cannot even rely on that, because your interaction could still have actually been just as one-sided as a comment that never received a reply from the person you are trying to talk to. A single click of a button to farm a person for more engagement - which is pretty transparently what this tool is designed for... Yeah, the ethics here are immensely messed up, no matter how you look at it.

[QwertyAD]

"It may be legal, but its not ethical."
Remember, google removed "dont be evil" from their mission statement.

[holedplot]

The internet isn't dying, it's been murdered.

[fusefurther]

I am so sick of Gen AI and every goddamn company trying to force it on us.
One of the most genuine parts of human life, interaction, taken away by a machine that has, unbeknownst to you, watched and documented your EVERY word.
That is horror novel stuff.
An opt in would be so easy. So incredibly easy but no. We never get a say in this :(

[d_trich]

It's good to know Google is ramping up their carbon output to create chatbots that people might not even use.

[Saintborntopeacesun]

We're getting closer and closer to the 'dead internet theory'

[cjkenney]

"just because we can, doesn't mean we should."

man why are people obsessed with ai writing things for you. how do people not see that this is your voice! especially with the ai writing OPINIONS for you. everyday this world feels like we are living in a dystopian nightmare.

[mozeytown]

When I was a kid, there was a family radio show I enjoyed listening to (yeah, I'm old) that invited fan mail and even had a segment where they read jokes submitted by listeners over the air. I sent them a lot of letters, hoping to hear at least one of my jokes on national radio, but in the process I also found out that the host replied to almost every letter (usually using a form response, but always personalized at least a little). This made me even more invested in writing in, as they remembered me from one letter to the next and made me feel valued. After a while, the show started using additional staff (people who never actually appeared on air) to help keep up with responding to fan mail, and I no longer felt like I was corresponding with the host. My interest dropped off pretty quickly and I stopped sending in letters. I lost all interest in the show not too long afterward.

The more things change, the more they stay the same.

[aequinoctiale]

eugh even the idea of this gives me genuine shivers. there's a reason why so many old stories have voice-stealing sirens and shapeshifters as a voice of evil. we as humans hate the idea of something else speaking for us with our own voice.