What These “Make Money With AI” Videos Don't Tell You — Part 1: Content Models

I've watched a lot of these videos now — “how I made $10k a month with AI,” “I built this in a weekend and it prints money,” that whole genre. I got skeptical early, so I actually tested a few of these myself and dug into the numbers behind others.

Here's the honest version: most of what these videos say is technically true. That's the frustrating part — it's not a lie, it's an omission. And which bit gets left out depends on whether you're coming at it as a developer or a marketer. A dev spots the technical holes instantly and misses the marketing ones. A marketer does the opposite.

To be clear — I'm not saying these don't work. People genuinely make money doing exactly what these videos describe. This is about what's not being said.

This is Part 1 of two, split by business model: content-driven ones here, platform-dependent ones in Part 2.

Amazon KDP: Is the Niche Already Saturated?

Here's the thing that struck me first, and it's not something these videos ever mention: a lot of the specific microniches being pitched right now are already saturated. Not “getting saturated eventually” — already tapped, recently, by the exact same AI-assisted strategy being taught. You can go check this yourself: look at the listing dates and sales rank on a “hot niche” in KDP right now, and a suspicious number of them appeared in the last few months, all doing the same formulaic thing. The gold rush isn't coming. For a lot of these niches, it already happened, and you weren't invited.

A few other things worth knowing before you sink a weekend into this:

• The royalty math is smaller than it looks. KDP pays either 35% or 70% depending on price tier, and print editions eat printing costs before you see a cent — per KDP's own published royalty-tier structure. The “profit per book” shown in these videos is usually best-case, not typical-case.
• New authors have to bootstrap their own visibility. Amazon's algorithm rewards books that already have reviews and sales velocity. Getting your first 10-20 reviews as a completely unknown author, with no existing audience, is genuinely hard — and it's the part these videos skip straight past.
• Amazon's actively fighting this exact strategy. There are real AI-content disclosure requirements now, and accounts can get suspended for flooding categories with obviously low-effort, mass-produced books. The platform you're building on is not neutral about the tactic being taught to you.
• You're only seeing the one that worked. The video shows the niche that sold. It doesn't show the same person trying the identical formula in forty other niches that sold nothing.

Affiliate Sites: Does Affiliate Marketing Still Work With AI?

Affiliate sites and AdSense sites share the same core problem, so I'll say it once here and it applies to both: this entire model runs on traffic, and traffic is slow. Not “slow if you're bad at it” — slow structurally, requiring consistent content over months before Google trusts you enough to send real visitors. Nothing in the pitch changes that; AI just makes it faster to produce the content while doing nothing to speed up the part that actually takes time.

On top of that:

• Google is actively cracking down on exactly this content type. Thin, AI-generated “top 10 best X” comparison pages are precisely what recent Helpful Content updates have been deindexing. You're being taught a strategy the platform you depend on is actively punishing.
• Commission rates are thinner than the pitch implies. Amazon Associates, for example, pays anywhere from 1% to 10% depending on category. “Traffic” alone doesn't translate into meaningful income — you need serious volume, and volume needs the traffic you don't have yet.
• Trust drives conversion, not just clicks. A reader has to actually believe your recommendation enough to click through and buy. AI-generated reviews with zero real product testing tend to convert badly, even on the rare occasion they do rank.
• The terms aren't yours to control. Affiliate programs can and do change commission structures, or cut affiliates off entirely, with no warning and no negotiation. You're building income on rules someone else can rewrite at will.

AdSense/Banner Sites: Is a Niche Site Still Profitable?

Same traffic problem as affiliate sites, with an extra wrinkle specific to this model: if you're building a single-purpose tool or calculator site to run banner ads on, you're not just fighting for traffic in general — you're going head-to-head with entrenched competitors who've had years, sometimes over a decade, to build up domain authority Google already trusts. AI can help you build the tool itself quickly. It does nothing to close that authority gap, which is the thing actually deciding who shows up on page one.

And the income numbers shown in these pitches deserve real scrutiny:

• Viewability inflates the pitch, not the reality. A banner ad generally has to actually be seen on screen before it counts as a real impression — plenty of “impressions” in a demo dashboard never really got looked at by a human at all.
• RPM (revenue per thousand views) is wildly oversold. Real RPMs are often a handful of cents, not dollars, and vary enormously by niche and where your visitors are located. A screenshot from one lucky, high-RPM month in a specific niche is not a representative baseline.
• Ad blockers quietly gut the real number further. A meaningful chunk of internet users run ad blockers. Even the impressions that do register in your dashboard don't all translate into ads a real person actually saw.
• The same Helpful Content risk applies here too. Fast, AI-assisted content built purely to host banner ads sits squarely in the category Google's been actively deindexing — you can build the site in a weekend and still lose it in an afternoon.

The pattern, so far

Notice what all three of these have in common: none of them are actually broken as ideas. Books, affiliate reviews, and ad-supported tools are all legitimate business models people have made real money from for years. What's missing from the pitch isn't the mechanism — it's the fact that every single one of them was, and still is, a distribution problem wearing a writing costume. AI made the writing part nearly free. It did absolutely nothing about the part that was always the actual bottleneck: getting someone to find it, trust it, and click.

Part 2 covers the other three models — YouTube, Etsy, and app stores — and a different kind of risk entirely: what happens when the platform you've built your entire income on top of decides, without asking you, to change its mind.

References

KDP Royalty Rates Explained: 35% vs 70% (2026) — profitable.app/kdp/stats/royalties — royalty tier structure and eligible price bands.