Today I’m launching a new section on my substack called Social Notes. This is where I’ll pen (sometimes lengthy) essays about what I believe social teams actually needs. I’ll share my pov and opinions, and you can take it with a grain of salt that these are just working theories. It’s informed by what I see in my current work at Slate and peppered with more than a decade working in social knowledge.

Before we dive into this ragebait topic, I should tell you where I sit before I tell you what I think. I run marketing at Slate, a content creation software for social media teams. I’ve spent an enormous amount of time looking at tools that promise to help brands create more content, including ours. When people complain about AI slop, they are complaining about an economy I participate in. I make money when teams buy tools. Weigh everything I say accordingly.

I have been posting on the internet since 2006, which is a decent amount of field research, if I do say so myself. I was definitely a user long before I made social a full time profession, and the question has always been the same: What can I share that another person might actually care about? After nearly twenty years of being chronically online, I have scrolled enough to be able to depict when things stop working because the content is just meh.

And lately, a lot of it does feel that way.

There is more content than ever, and a surprising amount of it seems to exist only to fulfill the requirement of being on the internet. It covers a slot on the calendar, giving someone a screenshot to paste into the weekly report for their boss that still refers to TikTok as the mint. It gets it’s 2 seconds of attention on the feed and then it disappears amongst the thousands of other posts created for roughly the same reason: to check a box.

Generative AI has become the obvious suspect and I totally understand why. Once you get duped by your first “Is this AI or not?” post, you develop a certain disgust and distrust. You start seeing it everywhere. You question the credibility of everything and everyone.

Is anything even real??

But most of what we consider as AI slop actually starts before anyone opens a prompt. It begins when a company decides it needs to show up everywhere, all the time, without giving the people doing the work more time or support. Publishing expectations have grown tremendously in the last couple of years. The teams have not. The pressure is more positioned on volume, and quality becomes something everyone still claims to care about and yet they structurally make it harder to achieve.

So that problem already existed. And when AI got introduced, everything just accelerated.

Across our conversations with prospects and customers at Slate, one problem comes up constantly:

We need to 10x our content without 10xing our team.

Ahhhhh, the classic “do more with less.”

It sounds like an efficiency problem, so, naturally, companies go looking for an efficiency tool. AI can generate a lot things. It can make the captions, draft the posts, create the visuals, turn someone into an avatar, and send that avatar out to say whatever needs saying that day. And it gets the job done for the most part. The content calendar is full and everyone has technically completed the assignment. Whether the work got better thought is a different question, and HubSpot has some points on this.

In their 2026 Social Media Trends Report, they found that 94% of social marketers now use AI somewhere in their workflow, yet 45% still call producing enough high-quality content their biggest challenge and 73% say it has never been harder to stand out. AI is doing exactly what was promised. It is helping teams produce more. But can we all agree that volume is the easy half of the problem??

The criticism around AI-generated content is not wrong. There is plenty of awful work being produced with it, and I have scrolled by and unfollowed many. But blaming the tool allows the people setting the expectations to get off fairly easily. We criticize the marketer who took the shortcut without considering why their team was expected to publish so much that they felt using a shortcut felt necessary.

I am not pointing at other people from a safe distance, either. I have shipped a calendar filler. There were times when I posted something that I knew was gonna be a complete bust while I was scheduling it, and it performed exactly like I had imagined it would—POORLY. My reason was that there wasn’t enough time to execute on something better. I was under-resourced, had little to no budget and simply did not have the time because leadership needed this launch announcement to go out today.

You see this firehose energy stems from the same sort of logic when a manager tells a social person to “make a post that goes viral.”

Based on what? For whom? Why would anyone care?

There is no strategy inside using generative AI other than needing to outsource a lot of the work to meet volume expectations and the under-resourced creators are left to manufacture the outcome by all means possible. When I see companies using AI to replace the creators, especially with things like AI avatar, I don’t know…it feels like the technological version of the same unrealistic demands of going viral. And if you ask me, I think it’s pretty gross.

You can’t replace a human with a machine. A person’s delivery and experience are usually the reason the content has any credibility. Remove the person and what? You’re now left with a synthetic cartoon shell of a person reciting a corporate brochure copy. I do not know why anyone would assume that would work. Also, I thought we were in the business of building that creative muscle. In order for you to get stronger, you need to have patience to build up some organic stamina.

Here is what technology companies keep getting wrong about creative people, and I say this as someone who works inside the industry—they assume creative people need help being creative.

They don’t.

The social people I know are not short on ideas. They are short on time. Their days are wasted rummaging through all the mundane things like organizing assets, sifting through feedback, chasing approvals, formatting content, and handling all the small production tasks that chips away an entire afternoon. Then they get given a tool that offers to generate 50 ideas for them.

Great. Now they have 50 more things to review.

Ask any good social marketer what they need help with and I’m fairly certain that it involves having some level of autonomy in the way they create. They know which moment deserves the cut and when a joke doesn’t land. They can tell when an overly polished piece needs to be killed. They are tapped into the culture and can tell you when you need something be paused. They can also look at a piece of content and decide “Nah, this is weird.”

And here’s the distinction that these ragebait slop conversation keeps missing. When people say AI slop, they are picturing only the output: the templatized generated captions, the avatar delivering robotic copy, the image that gave Steve an extra arm and a fourth finger. They are not picturing the mechanics underneath the work that have been incredibly useful for a while now. And nobody ever complains about them. You wanna know why they don’t complain?

Because it doesn’t change the end results that show up on the feeds.

Take webinars. We produce a lot of them, and I will be the first to admit that no one wants to watch an hour-long webinar twice. Especially not the social marketer who only needs the one banger line buried somewhere around the 20-minute mark.

Off the top of my head, here are two ways that I would go about with recycling this long form content into something more social-native:

• Transcript-based editing. This lets me read the conversation instead of scrubbing through it, find the promising section, and cut straight from the text. I can ctrl F and find the exact phrasing that I need.

• Then there are other tools, like Shortcuts, that’ll break the recording into a bunch of smaller clips and sort them by type. The Slate specific offer is that the clips are geared towards your brand and style of content that would resonate on your channels. But having this in my editor at least gives me somewhere to start. I can pull a single clip, combine multiple clips and edit it further.

These things all require AI. Nobody calls it slop, because we’re not just sharing work as is. There is further editing and tweaks but AI helped trimmed hours of work. That’s the intentionality behind creating content using AI. In this case, it never wrote a word the audience read. It cleared the drudgery that stood between me and the deliberate decisions to make great content. I don’t need to rely on AI to tell me whether the line is actually a banger, or whether the clip makes sense outside the full conversation, or whether it thinks my audience would actually care. I needed AI to sift through 60 mins of footage and surface those lines to me and then I decide what to do with them. And that’s the type of decisions that stay with me, and that is the correct place for it.

Slop is what happens when the tool crosses that line and starts making the calls. Technology is most useful when it is managed by the human.

Alright, I’m huffing and puffing a bit here but if you’re looking for some rule for how much of the work you should hand to AI. I can’t definitively tell you that. I can’t say that AI should handle 80% of your content execution, because execution contains dozens of small decisions that shape the final work. The better question is which decisions your team prefers not to outsource.

Creative discretion tops my list.

As I mentioned in my webinar example, a tool can surface a possible clip, and it remove repetitive work, or hand me a rough starting point. It should not be deciding what the I, or the brand, believes or what deserves to be shared.

Back to that one HubSpot report because it’s giving reassuring receipts. The three places where social marketers say AI is letting them down most are:

• image generation

• image editing

• and video generation

Notice the commonalities? These are the part of the work that’s closest to the final creative call. The best use of AI in a creative workflow is the mechanics of supporting you to work faster, not replace the creative thinking.

And by the way, a note for you leaders that are running social teams.

Creative judgement is not something you can easily get. Taste has to come from somewhere. You cannot ask someone to make culturally relevant work while giving them no time to participate in or at least pay attention to the culture, and a social team racing from one task request to the next gets fewer and fewer chances to use the instincts they were hired for.

AI can protect that time or it can eat it.

If a tool saves someone an hour and the hour goes toward improving the premise or reconsidering an edit, it helped. If the hour is instantly converted into a bigger publishing quota, the team is right back where it started. Which makes the entire resource kinda useless.

So why do we think this keeps happening?

In my opinion, part of why this keeps happening is that volume is easy to count and judgment is a lot harder to grade. A team graded on how often it posts will use AI to post more. The tool follows the incentive wherever it points. I would love to pin this entirely on managers, but the incentive does not only live above the team. Plenty of marketers chase volume on their own, because the feed requires it and gurus will tell you to be consistent which obviously means…post more. None of us are immune to this pressure.

Social teams do need help producing more. The demand is real, even when it is unreasonable, and pretending the technology has no value does nothing for the people responsible for getting the work done.

So here is what I would ask you to do, from one chronically online person to another. The next time somebody pitches you an AI tool, somebody will, and maybe it’s even me, ask what work it removes and what your people get back in exchange. If the answer something along the lines of produce more content and faster, fine. Just be honest that that was the goal. But if you want something quality then you need to plug in the resource that would buy back the time for your team.

Anyway. That is my read, and you get the gist of my thoughts on this ai slop convo. But if you want to hear other social pros debate about this in real time, we are bringing HubSpot to Slate’s next Social Social Club on August 12 at noon ET. Bryna Corcoran and Lauren Mendoza-Routt are joining Carmen Vicente to dig into how teams are actually using AI and which metrics better reflect the way social works. And if I may toot our own horn, we’ve built the SSC to be a pretty solid webinar series that leaves registrants looking forward to the next one.

Register here