FilmotionI can watch a beautifully produced video and learn nothing. I can also watch a rough screen recording that solves a problem I have had for weeks.
That difference matters more to me than whether either creator used AI.
As the founder of Filmotion, I work on a tool that turns ideas into short-form videos. That makes this distinction especially important: making a video easier to produce does not automatically make it worth watching.
I think of AI slop as content that occupies attention without giving much back. AI leverage uses the same technology to make a useful observation clearer, more accessible or easier to understand.
The difference is not the tool. It is what the creator brings to it.
Slop starts with something nobody needed to say
Imagine two videos from a local bike shop.
The first opens with, “Unlock your cycling potential with these essential maintenance tips.” It recommends checking the tyres, cleaning the chain and servicing the bike. The visuals look tidy. The advice is broadly sensible.
But I could replace the shop's name with almost any competitor's name and nothing would change.
The second starts with a close-up of a worn brake pad beside a new one. The mechanic points out the wear indicator, explains what it means on that particular pad and shows where a rider should look.
Both videos could use generated narration, assisted editing or automatic captions. Only the second gives the viewer a concrete reference they can use.
I call that leverage because the technology carries knowledge that already has value. In the first example, the technology mostly carries words.
I look for an information gain
My simplest quality question is: what does someone know after watching that they did not know before?
“Customer service matters” offers almost no information gain. A demonstration of how a shop handles a damaged delivery offers considerably more.
Specificity helps, but numbers alone do not rescue an empty claim. “Five secrets to better customer service” is still vague if the five secrets are listen, care, respond, improve and follow up.
I want the detail that changes someone's understanding. That might be the photograph a customer needs to send, the distinction between two similar products or the reason a common repair fails.
Before I develop an idea into a video, I try to finish this sentence: “After watching, someone should be able to…”
If my ending is “feel inspired” or “understand that quality matters,” I probably need to think harder. If it is “find the model number before ordering a replacement filter,” I have a useful starting point.
Evidence is the ingredient a prompt cannot replace
I can ask a model for a persuasive explanation. I cannot treat the resulting confidence as proof.
For a small business, the strongest material is often sitting nearby: a returned product, a frequently misunderstood invoice, a sample that aged badly or a customer question with an unexpectedly complicated answer.
Consider a furniture maker explaining why two oak tabletops look different. A generic script might talk about natural beauty and unique character. I would rather see the actual samples under the same light, with an explanation of the finish used on each.
AI could help organise that explanation or turn the maker's notes into a concise script. The samples still do the evidential work.
I also keep the limits visible. Two samples demonstrate those two samples. They do not establish how every oak table will look. Useful content can be narrow without being weak.
Polish can hide the missing work
The awkward thing about slop is that it can look finished before the thinking has started.
A confident voice, clean graphics and smooth transitions create an impression of authority. I try to judge the claim separately from that presentation.
I would test a proposed video by reading its script without the visuals or music. Does the explanation still hold up? Is there a reason to believe it? Does it answer the question raised at the beginning?
Then I would check the visuals. A generated image of a cracked pipe might illustrate the topic, but it cannot substantiate what caused a customer's actual leak.
For that claim, I would need evidence from the job and someone qualified to interpret it. When an illustration is useful, I want its role to be clear. It should not quietly impersonate documentation.
Leverage reduces the viewer's work too
I do not think production speed is the whole benefit of AI. The more interesting gain is how much effort I can save the person watching.
A dense product guide might contain the right answer but bury it across several pages. A clear visual comparison can make that answer easier to find, provided I check it against the guide and preserve important conditions.
For example, a retailer could show which of two chargers supports a particular laptop's stated requirements. I would name the models, verify the specifications and explain any relevant cable requirement rather than declare one charger universally better.
The result does not need to be groundbreaking. It needs to remove a real uncertainty without introducing a new falsehood.
My standard is something worth carrying
I am not interested in making creators prove how much manual effort went into every post. A useful explanation does not become less useful because software helped produce it.
But I do want a creator to stand behind what they publish. For me, that means knowing where the claims came from, checking that the examples support them and being willing to correct mistakes.
AI slop makes publication the achievement. AI leverage makes publication the delivery mechanism.
Before I open a video tool, I want something worth carrying: an observation, a demonstration, a tested explanation or a carefully bounded answer. Then faster production is genuinely useful, because it helps that substance reach someone who needs it.