YouTuber charged loads of fans $199 for shoddy machine-learning course that copy-pasted other people's GitHub code The AI hysteria has led to a rash of budding engineers hoping to land a cushy job somewhere in Silicon Valley. So it's no wonder that thousands flocked to an online course titled Make Money with Machine Learning fronted by Siraj Raval, a self-proclaimed AI educator, rapper, and entertainer with nearly 700,000 subscribers on …
I'd be more generous and say that he's fallen into the greatest cognitive trap of the crowdfunding era -- the idea that money comes first, then everything else falls into place.
Kickstarter was so swapped with this sort of thinking (for example all the non-engineers who said "give me some money and I'll hire an engineer to create my technologically impossible and/or financially infeasible games console) that they insisted on prototypes first (and now they all head to Indiegogo instead).
The notion of building a course around freely-available information is so much the mainstream that the notion on its own is valueless (how many of your uni lecturers invented and/or discovered the stuff they taught you) -- it's the execution that matters.
What I note as missing from the article is any discussion of quality assurance. If you're delivering a course to that many people, there should be several layers of oversight, and ideally also a fairly rigorous testing process involving teaching beginners... I doubt there was any of that.
It's the thing that bugged me most about Coursera and Udacity when they first came out -- 1000s of people taking courses that had never been run before, and many of which (on Coursera and EdX, anyway) were only ever run once. No beta testing, no refinement or improvement... all people ever saw was the first draft. In that situation, at least these courses were based heavily on tried and tested university modules, but even then, the change of medium really called for a lot more in terms of adaptation and testing.
This post has been deleted by its author
From what I have seen, courses are successful when they act as a convenient training program which helps users get started with a technology/domain with minimum effort. There are already a lot of information available and easily accessible, in the form of books and papers (many important papers have an archived version, but not all of them have one- thus, some of them are completely free). But the number of people who will directly benefit from such sources are very less.
The small group of people who benefit from these primary sources will have to make the material even more accessible, to reach a larger audience who finds it hard, or cannot learn directly from research works. This goes on, as new groups of people work towards simplifying the material even further. And finally this reaches the majority who rarely benefit from reading papers or reference books. Tutorials target the lowest level and are supposed to be as easy as possible, enabling practitioners in the field to easily adapt to the new way of doing things.
Thus, as long as the instructors break down the material quite well and explain everything step by step without having the users do a lot of work, the course should become a success. Yes, I do agree that there must be some level of beta testing. But there are users who would volunteer to pay and sign up for new courses that are marked as "beta", as long as it covers everything they need and they are given a trial period to evaluate if the course will really benefit them.
But the course mentioned in the above article is quite suspicious, if whatever that was said was true.
All very much true.
However, the good course designer starts with a particular demographic in mind as his/her target audience, and builds the course around them. A good teacher will adapt the course on the fly if students are finding it too hard or too easy.
Modern digital courseware wants to sell to as wide an audience as possible, which means all notions of prerequisite learning go out the window. Then there's the tendency for everything to be live coding instead of lectures, which means everything's paced by lines of code rather than complexity of concepts.
I've been watching a LinkedIn Learning course on Vue.js today. and it's exclusively live coding. The presenter keeps going off on tangents just in case you don't know particular features of JavaScript as and when they pop up in a line of code he's writing.
..Eli The Computer Guy. Originally thought he seemed OK with his free courses on YouTube. Then noticed that most of his "Courses" consisted of one or two videos on the subject, then he'd move on to something else. Got the impression he was just taking the first two chapters of a book and blurting them out in the hope to get viewers and subs (unfortunately has worked). I gave up on him when I watched his backups video. The amount of disinformation in that was clear he has never implemented a proper backup solution before.
Yeah although i can understand why they want to be. They then feel like they are working for themselves and not some faceless corp. I find it hard to explain as you'll get some that want to do it for that reason and will shit on anyone that gets in the way or misunderstand the approach. Such as the guy that started up a video podcast for Star Wars Republic as it was being released. Was doing it from his house. Which was all fine. But then he thought to be bigger he needs to go and rent an office space with big TVs in the background. Long story short, didn't last long after the move and "hiring staff". That's where a lot of them go wrong.
We all want to work for ourselves and YouTube used to be a way to do it. But I think getting lucky with it so you get the numbers that you can now do it as your day job but ignore the bullshit "YouTube celebrity" side of it is just fine. Its why I like the likes of LGR. He's done it as a hobby for years and now finally made it a success & was able to quit his job a while ago. But he doesn't get involved in the "celebrity" side of it. He's even admitted in an old video he wouldn't do meet ups with viewers he doesn't know. As its really uncomfortable meeting strangers as it is. Strangers that think they know you just cause they watch you all the time must be even weirder.
a self-proclaimed AI educator, rapper ..."
.And I stopped reading at that point. No more info about this guy needed.
Millennials - if you want to learn something save yourself a dumb online fee and instead try reading up on it (even - gasp - a book or 2) then putting what you've learnt into practice. Its worked for millenia (ironically).
As this comment is not accepting votes. vote here
In order to "make money with machine learning", the tool you need to know how to use is Powerpoint, and you need to use it to persuade people who have money to part with that money.
For example, Andrea Rossi managed to persuade Woodford Capital to part with £32m for his cold fusion "invention", and there is very little evidence that it actually works.
He's a rank fraud and nothing else.
He claimed to have authored a best-seller on his website. That book on Amazon has a paltry 3 star rating from 17 ratings and multiple reviewers have shared their suspicion of manipulated 5 star reviews.
See:
https://www.amazon.com/Decentralized-Applications-Harnessing-Blockchain-Technology/dp/1491924543
Besides this, he removed all references to Udacity from his LinkedIn profile, which is extremely suspicious given his deep involvement in their deep learning nanodegree programs for at least 2 years now.
https://www.linkedin.com/in/sirajraval/
Someone I know who has been thinking about getting into AI pointed me at it. I noted right off the top that there were not one but three typos in the ad, and that the reference article was written by the person giving the course. Both red flags in my book.
I am currently a software engineer in a data science team producing software that yields millions of dollars in revenue for our company. I did my undergraduate in physics and my professors encouraged us to view MIT Open Courseware lectures alongside their subpar teaching. I learned more from those online lectures than I ever could in those expensive classes. I paid tens of thousands of dollars for that education. I decided that it was better bang for my buck to learn data science than in would every be to continue on in the weak education system we have globally. I paid 30 dollars month, for a year, to pick up the skills to get into data science. I landed a great job, paying a great salary because I took advantage of these types of opportunities. If you hate on this guy for collecting code that is open to the public and creating huge value from it, then you can go get your masters degree for $50-100k and work for someone who took advantage of these types of offerings. Anyone who hates on this is part of an old school, suppressive system that will continue to hold talented people down. Buck the system and keep learning!
Your comment starts off reasonable. We have all seen subpar teaching in established universities, and we all know there are very good online teaching materials. I'll gladly agree to that.
And you then turn around and say that, because this is the case, all the material online is better than a university. That's just wrong. The course referred to in this article, for example, started off pretty badly in that it didn't teach the concepts people need to learn. And this guy doesn't get credit for useful code someone else wrote either; I'm fine if he chooses to teach from it, but he should properly credit the original authors and choosing useful Github projects does not a good course make. The internet has a bunch of information, and for every very helpful resource out there, there are at least ten pages with something outdated, incorrect, biased, or useless. Your all-or-nothing stance is misguided.
As a Novell-certified engineer, I remember well the number of "paper CNEs" who entered the field in the 1990's. These folks had no idea which side of the floppy disk to stick in the drive, but they studied the book and passed the exams. No doubt many of these went on to "teach" their newfound "skills" to others for a fee. My mother held a doctorate in Education, so I learned at a rather young age just what was involved in "educating" students, and it was/is far more than as presented in this article. Still, people are attracted by a low price for learning (no shame in that, but as has already been said, buy a book or two: it's even cheaper).
Also, since when does the number of github-hosted repositories denote someone's technical abilities beyond the ability to click the "fork" button repeatedly?
All of this aside, caveat emptor. If you're going to hand over $200 to someone you don't know, read the fine print. Look for the satisfaction assurance, and if missing, ask before purchasing. Heck, I don't even sign up for forum access without reading the privacy policy and ToS (no, not Star Trek, gang!). I know I'm in the minority concerning my level of retentiveness, but due diligence is mine to perform alone. If I fail, I have only myself to blame.
When I visited this "AI Experts" Web Site I was surprised to find it almost impossible to even scroll down the page, I am not currently running an AdBlocker and am sure if I was then I would not have encountered this 100% CPU in my Browser.
It appear's he's running Browser CPU Cryptocurrency Mining on his Web Site?
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
