I binged 13 AI courses from Udemy and Coursera (updated)

Hey Innovator,

Three months ago a new cohort of students joined the Innovating with AI Incubator – but not without first considering all the AI education options.

Over the past two weeks you saw 8 of them present their brand new AI prototypes at Demo Day (replays here).

They turned their AI ideas into reality in less than 60 days and become part a community of like-minded creators. At IWAI, we're building a community of doers – that is, people who take action rather than just passively reading about AI.

Here's how Dr. Grady Batchelor, one of our Incubator students, described it during Demo Day:

This was one of the huge discoveries I made when I started binging all the cheap AI courses I could find (more on that in a moment). They were all talk – and talk is cheap!

The Innovating with AI difference:

• We don't just talk – we show you exactly how to build your 10-hour prototype
• Then we give you a serious but plausible schedule to get it done
• Then we help you launch it to 100k+ people at Demo Day

But I know that to make sure IWAI is the absolute best it can be, I have to continuously keep an eye on everything else that's out there.

So over the past few months I've been buying courses and certifications on AI – like, actually going through them, to put myself in your shoes.

I wanted to answer this question.

“If I were an IWAI reader, and I went to look at AI courses on Udemy, Coursera, and even YouTube, what would I find?”

If that sounds even a little interesting, then grab your popcorn my friend 🍿, and get ready to enjoy the show 😎

Part 1: Udemy 🤑

When I asked tens of thousands of readers what they would do if IWAI didn’t exist, one of the most common responses was: “I’d probably buy one of those $10 courses on Udemy.”

I’ve browsed Udemy before, and even posted one of my freelancing lessons there many years ago. But to be honest, I had never really dug into the content until recently. When I’d invested in professional development and online courses in the past, I’d always gravitated toward the really deep stuff – hour-long lessons full of actionable advice and practical examples, like Ramit Sethi’s Earnable, which helped me ramp up my web dev business when I was just starting out.

On Udemy, the content is much more “bite-sized.” In fact, I found the same trend on a lot of other platforms too – LinkedIn Learning, Coursera and others. It seems like they’ve all been TikTok-ified – almost everything is a 3-minute video and it’s really hard to focus for long enough to get anything useful out of it.

That said, I would be lying if I didn’t admit that “a $10 course” seems like an attractive proposition. But at the same time, something about that extremely low price tag just doesn’t add up. How can the economics of this really work?

I did some more research and learned that... the economics don’t really work. Even the best instructors barely get paid ... which means they have zero time to engage with their students, keep their content up to date, or work with you directly to ensure you have everything you need to succeed.

In fact, Udemy’s terms restrict how your instructor can communicate with you – for example, your instructor isn’t allowed to see your email address (it’s Udemy property). They’re also not allowed to post an external link to a survey about your experience – they’re often even prohibited from linking you to their own website.

We would never have been able to build the IWAI Incubator without rapid, timely, large-scale student feedback. Everything we do is based on interviews and survey responses from thousands of AI enthusiasts. Udemy’s rules prevent instructors from doing the #1 most important thing – collecting honest and detailed data to improve their curriculum.

Imagine, in a field that’s moving as rapidly as AI, telling an instructor that they are prohibited from getting in-depth feedback from their students if it requires the student to click an “external link.” Not exactly a recipe for innovation.

What are you learning for $9.99?

Let’s look at one popular AI course on Udemy, “ChatGPT & Midjourney: 23 Ways of Earning Money with AI.” On the surface, this seems like something that an IWAI reader would like – after all, many of us are entrepreneurs or executives who see the potential in AI and want to level up our careers or businesses.

Once you dig in, though, it turns out the course is almost entirely about making money on Fiverr, the platform where people do things for $5. In fact, there’s even a significant subset of content that’s about using ChatGPT to create courses about ChatGPT... which is concerning to say the least.

Here is the basic strategy they present across “72 lectures” and “7 total hours”:

1. Set up a Fiverr account
2. Advertise that you will do something for $100, such as “I will write perfect ad copy for your facebook ads” (direct quote including lack of capitalization)
3. Wait for a client to hire you
4. When the client hires you to do the thing, you just do it with ChatGPT and send them the results, and they pay you for the ChatGPT output.

Direct quote from Lesson 22 at 5:44 –

😑

What is happening here?

First, the business models presented here only work if you are living in a country with a very low cost of living and performing geoarbitrage, which was popularized in the early 2000s when people in the United States and Europe realized they could pay people in India and the Philippines very low wages to do computer tasks because of the cost-of-living disparity between the countries.

For example, the average monthly salary in the Philippines is $329 USD. So if you can bank a handful of these ChatGPT-copy-and-paste deals every month, you’re golden. (Fiverr takes 20%, though.)

The weird twist here is that ChatGPT is actually the thing that makes geoarbitrage completely unnecessary. For example, even if a client wanted to use Fiverr to write a Facebook ad, how long would it take them to realize that they could just type in the prompt into ChatGPT themselves? (They get a crappy ad either way, but ChatGPT gives them an even cheaper crappy ad than a freelancer does.)

The course purports to teach you how to create a low-end Fiverr business but in fact is an accidental case study about why these Fiverr gigs are going to cease to exist soon.

That said, there is one area where I have found that OpenAI tools are really helpful with creating ads – using DALL-E 3 to create compelling imagery for ads. This has allowed me to move much faster and stop using cheap freelance design services. My higher-end designer now just works on really complex user experience and branding projects, and she doesn’t need to do things like “create 9 variations of a guy looking really happy at a computer” since that can now be done via DALL-E.

In other words, every good scam needs a grain of truth. And that’s exactly what you get in the $10 course that claims to show you “23 Ways of Earning Money with AI.”

Udemy’s very awkward fake-pricing problem

The “23 Ways” course supposedly has 9,000 students and 1,000 reviews averaging 4.6 stars, including gems like “This is amazing. I will apply all your lessons to create businesses with AI. Thank you so much for the quality of your course.”

So, why are people buying this stuff?

Well, part of the reason is that there’s an elaborate marketing scheme happening on top of the “make money with AI” scheme.

In 2022, Udemy paid a $4 million settlement to end a class-action lawsuit that accused the company of deceptive pricing.

The nature of the alleged deception? Their courses are nearly always advertised with a very high “strikethrough” price and a very low “actual” price, like this:

Why would this be illegal? Well, these courses are almost always promoted at an 80%+ discount. In California and many other states, you are not allowed to create fictitious markup prices to make products look like a better deal than they really are. There are legal thresholds that require you to actually sell the product at the “original” price a certain amount of the time. JC Penney and other retailers have paid $50 million+ to settle brick-and-mortar cases alleging similar deceptive behavior.

The out-of-court settlements allow the retailers (including Udemy) to avoid a jury trial and deny wrongdoing. In fact, in Udemy’s 2022 annual report, they bragged that the $4 million settlement was “immaterial” – in other words, the lawsuit is just a small cost of doing business and they plan to continue doing things that their own customers say are deceptive.

Let’s think about this for a moment.

We have a course that is truly worth about $9.99 – less than the cost a movie ticket and bordering on the price of a Happy Meal.

But Udemy knows that if they admit the true value is a measly 10 bucks, people will realize the course must be low quality and avoid it. To generate more sales, they add a superficial anchor price of $94.99 (which nearly zero people actually pay) in the hopes that customers will think the course is 10x better than it actually is.

It’s as if a Motel 6 advertised rooms for $57 but said, “don’t worry, they’re normally $700!” We’d all laugh and slowly back away.

Instructors that can’t afford to talk to you

The self-destructive economics of cheap-o pricing also make it impossible for educators to take their curriculum (and students) seriously. In the case of Udemy, the instructor only gets paid 37% of what the student spends – Udemy gets the other 63% as its platform commission.

On a $9.99 purchase, the instructor gets $3.70. 🤯 (You tip your DoorDash driver more than that.)

Honestly, when you think about the guy making $3.70 to tell you to copy and paste things from ChatGPT and sell them on Fiverr, it’s all kind of sad.

But it gets worse. Udemy’s own “Marketplace Insights” tool tells instructors that the “top monthly revenue” to expect from an Artificial Intelligence course is $8,001 before Udemy takes its 67% cut, which means the most popular instructors can expect about $2,961 a month in income.

And that’s for the best performers. Half of all AI courses earn less than $31 per month!

To put it another way, the manager at your local McDonald’s makes way more than your Udemy teacher. At LinkedIn Learning, instructors are excited to make $600 a month.

So how can you possibly expect those instructors to treat their courses and their students with the time, attention and care required to succeed in a rapidly changing industry like AI?

They can’t afford to keep their courses up to date.

They can’t afford to build a community.

They can’t even afford to talk to their students.

Case in point: The guy with the “23 ways to make money” course knows he should be keeping his material up to date so he can get a “Recently Updated” badge on Udemy. But so far, he has only managed to add 8 minutes and 19 seconds of new “lectures” since May –

The students lose. The instructors lose. The only winner is the platform, which takes home the lion’s share of the money while doing nearly zero to help anyone learn or succeed.

Part 2: Coursera 😴

Coursera is another popular choice for inexpensive learning – this time with the twist that they partner with some Big Tech companies and universities to be the delivery platform for those organizations’ trainings. Coursera’s business model is kind of opaque – almost everything is publicly listed as free, and you never really know what the price is until you click the “try it now” buttons.

I went ahead and clicked the buttons and started a lot of free trials.

First and foremost, Coursera is a legitimate platform for getting technical certifications for your job, particularly if you are already a software developer. For example, you can get a Meta Front-End Developer certification there, and it’s real.

However, that’s really not what we are focused on here at Innovating with AI. The industry is moving so fast that spending months learning to code or getting certified by Microsoft should be items #998 and #999 on your to-do list. You’d only benefit from these programs if you already have the skills and just want an official certification to prove it. These certification courses and corporate trainings are not where the real learning happens.

The other thing about corporate trainings – they’re deeply, irredeemably boring. In fact, the one thing that seems to unify all the lessons on Coursera, including the ones that don’t come with a brand-name certification, is that they are the blandest and slowest experiences you can possibly imagine.

For example, there’s a group of instructors by the name of DeepLearning.ai that have produced a ton of Coursera material. (Why so much? See the “we get paid pennies so we have to be a content farm” section above.) Most of their stuff is developer-focused (“get your TensorFlow certification”), but I bought and took one of their general interest courses, “AI for Everyone.”

Not only were the videos painfully slow and boring, they also seemed to diligently avoid saying anything that might result in the student gaining hands-on skills or doing anything in the real world. For example, here’s the diagram they provide to help you brainstorm AI ideas:

Breaking news: you should figure out what is “valuable for your business” ... 😑

If I had to guess, I’d say the instructors take this approach because they want their training to be low-intensity, low-controversy, and basically harmless. That way, anyone can take it and say they “did an AI course,” and there’s no scenario where someone might be pushed outside their comfort zone or forced to change their behavior, build something, or learn something new.

That approach allows you to sell lots of corporate trainings that people subsequently sleep through. And it results in slides like this one, which appears in Week 3...

I should... tell my friends to take this course? Start a reading group?! Hire a few machine learning and data science people to help?!?

Yikes.

Before we wrap, I want to share one cool side note about Coursera. You can actually get a real Ivy League master’s degree on the platform! (It costs $50,000, though.) In the same way that tech companies use the Coursera platform to deliver their (albeit mind-numbing) certification courses, some universities do the same with full degree programs. It’s an exciting innovation and really worth doing if you’re in the market for a degree over the next few years.

But right now, our focus is on taking action, not spending three years on a master’s degree. So let’s dive into another free learning option...

Part 3: YouTube 😵‍💫

I learn a lot on YouTube. For example, how to remove a fence post from my chicken coop with a clamp and two 2x4s and how to beat the final boss in The Legend of Zelda: Tears of the Kingdom ⚔️. Some of my favorite podcasts repost their episodes on YouTube, and my 4th-grader is always game to watch endless hours of teen influencers pranking each other.

There’s even some good starter material on AI, like Wharton Business School professor Ethan Mollick’s five-part series (about an hour in total) called Practical AI for Instructors and Students.

When I hear people say “I’ve learned a lot on YouTube,” I think it’s stuff like Mollick’s videos that they’re referring to.

But is YouTube really a learning platform – or is it a fun way to kill time?

First, there are the ads. The typical free YouTube viewer is interrupted by an ad about every 5 minutes, which, as you might imagine, makes focus and deep work nearly impossible. (I pay for YouTube Premium so my son doesn’t have to deal with ads while watching his prank videos.)

Then, there’s the algorithm. Remember that YouTube is social media, just like Instagram and TikTok, and its sole purpose is to keep you mindlessly clicking through a wide variety of dopamine-inducing short-form content. Even if you want videos like Mollick’s, you’re going to get videos like Corgis Locked in a Bitter Argument - Who Will Win?! (They know I can’t resist a corgi.)

There’s just no way around the fact that the entire purpose of YouTube is to distract and addict you. Yes, you can cherry-pick useful videos, but the moment you try to sit down and use the platform for an extended period of time, you’re sucked into the vortex.

Even Mollick’s YouTube lessons, which are legitimately worth watching, are forced to acknowledge the reality that YouTube is a destroyer of attention spans — it’s an hour of learning (very reasonable) but it’s split into 5 parts because… apparently nobody on YouTube can watch something that’s more than 11 minutes long.

As great as it is to find a helpful YouTube video now and again, I’d much prefer an education that’s not dictated by a social media algorithm.

What’s missing?

As I look back on my $0-$10 course binge, I’m struck by how social media and internet addiction have completely derailed the process of learning a new skill.

Everything is short and TikTok-ified, to the point where even Ivy League professors can only communicate in bite-size chunks.

Nobody gets paid an honest wage for their time, so at best, it’s a side hobby, and at worst, instructors are pressured to build “content farms” where they focus on quantity of 3-minute videos over quality of education.

There is ZERO community (unless you count YouTube comments filled with psychopaths 😭). And there is ZERO opportunity to actually meet your teacher, work with them personally, and build a relationship that helps you learn and grow.

It all adds up to a frazzled, passive experience, and I suspect that the end result is that students glance at a few videos and then give up and move on. After all, if you’re not investing anything and not getting any personal help and everything is sliced and diced for the attention span of a goldfish… why even bother?

Honestly, I didn’t even know how bad it was out there until my students told me. There is so much static in the online learning world right now that it’s easy to forget what really matters:

• A curriculum built for focus, deep dives into serious topics, and real-world projects that cement the knowledge for the long term.
  ​
• A community of enthusiastic, like-minded fellow learners — so you can build your network and help each other toward common goals.
  ​
• A coach and mentor who’s actually there for you — and who’s constantly working on adding new material and keeping the program up to date.

I didn’t intentionally build the Innovating with AI Incubator program as a counterpoint to the social-media-frenzy of cheap online courses. But now that I’ve seen what’s out there, I’m even more excited about the Incubator and the hundreds of students we’ve already helped start their AI journeys.

If you haven't watched the replays of both April Demo Days, click here to see what 8 Incubator students built in 60 days. And tune in on Tuesday for our third Demo Day, featuring four more Incubator student launches.

Next week, we’re opening the new cohort of the Incubator program and offering a huge bonus for everyone who joins on Opening Day. If you’re interested in learning more about the Incubator, click here to get on the waitlist via text message or WhatsApp.

And as always, I’m here to help if you have any questions — just hit reply.

Till next time,

– Rob

​

PS. The new cohort of the Innovating with AI Incubator opens next week with a special Opening Day bonus. Join the waitlist via text message or WhatsApp to make sure you're the first to know.