The Bubble We Built Ourselves

One of the most talked about questions I encounter in this space has always been around something like: "Do my Subs and Views-counts reflect my level of talent?" or "What am I doing wrong?" when looking at some random AI slop sitting at 250k views on YouTube. Well, for the most part, I always knew the answer and I want to be honest about this stuff. Not to be negative, but because I think staying comfortable with a comfortable lie is worse than the discomfort of a useful truth.

"Am I good?" has always been the wrong question. We (meaning the content creators using music made through and with AI) have an ecosystem problem. And I think, at least partly, we built it ourselves.

Look around any AI music community today - Discord servers, Facebook groups, Reddit threads, YouTube comment sections. The overwhelming majority of people engaging with our content are other AI music creators. Artists supporting artists. Creators encouraging creators. It feels good. It feels like community. And it is community. But to me, these communities are also...well ...a trap of sorts... one we've been happily walking into every week, not realizing that every supportive comment, every premiere watch, every shared post, was quietly teaching these platforms something about us that's actually very hard to undo.

I ran an experiment. I'm going to tell you about it. And then I'm going to be honest about what it means for our communities and things like Fresh Friday Festival, an event I genuinely love and help organize, but one (amongst others) I probably see with different eyes as the rest of you.

How the machine actually works

The Algorithm Doesn't Watch Your Videos

First things first...here's the thing you need to understand about YouTube, Instagram, and Meta's recommendation systems: the algorithm doesn't evaluate your content and decide whether it's good. It watches who watches you - and then uses that information to find more people like em.

This is called collaborative filtering. The system groups users together into clusters based on their shared behavior... what they watch, how long they stay, what they watch immediately after. When you get a view, YouTube doesn't just count it. It logs the viewer's entire profile. Their watch history, their interests, their subscriptions. And it asks: "Who is the type of person that watches this channel?" Then it finds more people who match that profile and puts your content in front of them.

The implication is enormous. The algorithm isn't building a model of your content. It's building a model of your audience.. It always has, Google at it's core is serving Ads and selling Ad inventory. Meaning their entire Business model is build around understanding Audiences. So it goes looking for more of your audience....Meta and all the other company's too for that matter. Not for altruistic reasons mind you. Every creator to them has the potential to gather crowds of watchers consuming your content where potential Ads could be seen, they see you as just another reason that people watching you stay an their app/service for another few minutes. They have every incentive to make your channel succeed. So when you see the view count on your profiles don't ask yourself "does YouTube hate me?" or "is my AI music video good enough?" The real question has always been: "Who is YouTube learning to show it to?"

The deeper problem

We Trained It to Find More of Us

Here's where it gets a bit uncomfortable. Every time we post our AI music videos in our creator communities, every time we share in the Discord and ask people to watch, every time we gather a group of creators to view each other's premieres we are feeding the algorithm a very specific audience signal. We're teaching it, systematically, that our content is for other creators.

Not for music fans. Not for people who want something to listen to while they work. Not for the person who types "dark cinematic orchestral metalcore" into the search bar at 11pm. For creators.

And here's the cruel logic of collaborative filtering: once the algorithm has built a strong model of who your audience is, it uses that model to find more people like them. If your audience is 80% creators, it will keep finding creators. According to a technical breakdown of YouTube's recommendation architecture, user profiles are continuously refined in real time as new signals come in, meaning every creator-heavy viewing session deepens the bias, not just creates it.

This is the loop. And it's self-reinforcing. Once you're in it, every new creator-community view deepens the pattern. The algorithm becomes increasingly confident that your channel is for creators - and increasingly unlikely to suggest you to anyone else. It still can happen, but it's very very hard.

The genre problem compounds it

Jumping Genres is Signal Pollution

Now, there's another layer to all this, and I see it constantly in our communities. A lot of AI music creators - myself included at times - love experimenting. Hip-Hop one week. Some crazy EDM the next. Something random the week after. The AI tools make it incredibly easy to explore, and that freedom is genuinely exciting.

But from an algorithm standpoint, genre-hopping is an absolute disaster. Here's why: the algorithm isn't just tracking who watches you. It's also building a model of what kind of person your audience is, and what other content they enjoy. When you jump between genres, each video attracts a completely different audience segment and those different audiences send conflicting signals about who the channel is "for". (Again let me stress this: it's all about audiences with these company's)

Instagram made this explicit in their December 2025 algorithm update. According to an analysis of the update by Net Influencer, the platform now explicitly states that accounts which publish across unrelated themes "often experience weaker audience matching and less consistent distribution." The algorithm uses a rolling window of your recent posts to classify what your channel is about. If that window contains five different genres, it can't confidently put you in front of any audience, because it doesn't know who you're for.

YouTube's Creator Liaison Rene Ritchie put it simply: a YouTube channel that makes random videos is like a restaurant that serves random food. It's technically fine. In practice, impossible to recommend with confidence.

What I found out

The Silent Channel Experiment

About five months ago, I started a second YouTube channel. I told nobody about it, not my community, not anyone from Fresh Friday Festival, not the villains, not a single creator. It remains anonymous and will forever stay that way.

But the rules I set for myself were simple: post every 8–12 days. Same visual style each time, same general sonic territory. No cross-promotion. No community shares. Just consistent, on-brand content, posted quietly, and left to find its own audience.

The growth wasn't viral. But the trend line told a story. After a slow start where the algorithm was learning what the channel was, views began climbing steadily and they kept climbing. By February 2026, the channel was routinely getting 1,500+ daily views, 500 watch hours done...all this completely organically. I didn't lift a finger but creating a video every other week.

Compare that to my main channel, where a full year of community support, premiere events, and creator network cross-promotion have produced ....nothing really, subs are stagnating, i don't get views on my videos, even though the quality of content is much much higher. The difference? Who was watching first, and what the algorithm learned from them.

The algorithm on the silent channel was finding real music listeners. People who don't make AI music. People who just wanted something to put on. The kind of audience that, once it finds a channel it likes, actually comes back.

The hard conversation

What This Means for Fresh Friday Festival

I need to be honest here because I'm one of the people who helps run this thing.

Fresh Friday Festival as it stands is mostly a community watch party, creators supporting creators, it is something I genuinely believe in as a community event. The connections are real. The encouragement matters. Finding out someone half way cross the world is making the same kind of music you are and actually giving a damn about your work...that's not nothing. All Drama aside... That's actually meaningful.

But we have been telling ourselves, or at least implying, that the event also provides a meaningful algorithmic boost to participating creators. And I don't think that's true. I think it might, in some cases, be the opposite of true.

Here is the blunt mechanical reality. When a group of AI music creators gathers to watch each other's premieres for 10-minute slots, YouTube observes a group of viewers who are: highly active content creators themselves, with short watch sessions, low per-channel catalog depth, and low return rates. That pattern does not look like a devoted audience discovering a new artist. It looks like creators doing a promotional circuit.

This isn't anyone's fault. It's just what the data looks like from the outside. YouTube's 2025 algorithm update specifically moved toward judging channels as a whole, prioritising patterns over individual videos - meaning a sustained pattern of creator-only engagement isn't just bad for one video; it's gradually shaping the identity of the entire channel.

What we should do instead

Rethinking the Strategy

None of this is an argument for abandoning the community. The community is real and it matters. But I think we need to separate two things we've conflated: community support as a social good and community support as a growth strategy. The first is genuinely valuable. The second is, based on everything I can find, largely a story we've been telling ourselves.

Fresh Friday Festival should probably continue, as a community gathering, a place to connect, a way to maintain the human fabric of something that can otherwise feel quite isolated. This is a stand in obviously, you can exchange FFF with any other streaming show/event. But we should stop describing these events as an algorithm boost mechanism. It isn't one. And pretending it is sets up creators for confusion when their numbers don't grow despite consistent participation.

OK so here are some of the things that the research actually supports:

The actual point

The Bubble Is Ours to Break

Let me get one thing straight, I'm not arguing against communities. I write this as someone who is sitting right inside this bubble. Not someone looking at it from outside. My main channel has the same problem I've described. My early videos were shared almost exclusively on Reddit of all places. My initial audience was almost entirely other Suno creators and ...well AI haters i suppose (thanks Reddit). The algorithms learned what they learned, and that's that. The channel was always predominantly done by me for me and I suppose anyone in the community that likes what I do, i don't care to much about subs or monetarization, i have different avenues for that. This is at it's core... a hobby for me. A hobby that's also a passion, one i take ...at times...probably more serious than I should.

The silent channel experiment also didn't make me feel smarter. If anything it just confirmed all my suspicions. One I expressed to a few creators before, some of which asked the very questions in the beginning of this article. The answer was always uncomfortable and I'm not sure if they listened or wanted to. I still wager, you - reading all this, will rather be uncomfortable and in the known, than comfortable and still feeling like you are not good enough.

The talent is real in this community. Some of the work being made right now is extraordinary and deserves a bigger stage. The tragedy is, that it will only exist in these communities and rooms of people who already know how to make it. But if all this, in the end, is also just for yourself, for fun, for relaxation, for therapy ...name the reason... for just scratching that musical itch. If you never even planned to make this into anything else but personal enjoyment, more power to you. All this I just wrote won't matter to you. Don't worry about it. Enjoy being creative!

As for the rest: If you read this...here is my gift for you today. The gift of understanding.
And understanding why you're invisible to everyone else, is the first step to not being invisible anymore. But take solace to the fact, that all this, most likely, has nothing to do with your talent as an artist. Or maybe it does, but not in the way you thought of it before.
Anyhow, just do me a favor, don't shoot the messenger. I'm getting enough flag these days for sticking to my guns 😅... but f*ck it, this is the whole reason I build all this in the first place.
Now, if you have any questions, under this article you can log in with your Google Account and drop a comment, DMs are also open.

Cheers,
Aidan