The Rise and Fall of Siraj Raval

Source: Instagram / montage

The world of Artificial Intelligence and Machine Learning(AI/ML) is relative drama free, especially when you compare it to the world of crypto (where there's drama happening almost every week). But recently Siraj Raval, the biggest YouTuber in this area fell from grace, and he fell hard. Siraj was (and probably still is) on a mission to educate the world about AI and ML and how to use it to "make the world a better place" (to use some Silicon Valley lingo).

The Rise

Both, Artificial Intelligence and Machine Learning are rather old disciplines, their roots go all the way back to 1956. AI in particular had some hype-cycles in the past, which were followed by an AI Winter. Outside the hype-cycles, AI/ML were rather academic disciplines and boring for most people who are not involved with it. Until Siraj came. He managed to explain the concepts behind AI/ML and do it in an entertaining way. He is also excellent at grabbing people's attention on social media, a must in today's world. He started uploading videos in 2016, interestingly he started with videos about Bitcoin, but switched to AI/ML videos about three months later. Looking back at his rough early videos, you can really see how he improved his craft of making videos, I guess he also improved his skills in social media marketing. Combine that with more and more people are getting interested in AI/ML (probably seeing this as an opportunity to make a lot of money) and you have the perfect recipe for getting a huge following on YouTube (currently around 690.000 subscribers) in a relative niche topic, which is not very accessible (linear algebra and statistics are a must in this field) and rather boring at times (I work in this field and really like it, but I can't deny that the fundamentals are rather dry, and again: you need them).

I stumbled upon his videos in 2016 on Google+ (remember that kids?) and I really enjoyed them, there is something about learning difficult topics in an entertaining and fun way, I also liked the topic he was talking about, because they were rather cutting-edge at the time (they are still cutting-edge today). He was also the first YouTuber I ever subscribed to (and still is one of the few, I'm subscribed to). First I thought, that this is niche content, only interesting for a few people, who are like me. But looking at his subscriber count showed, that this assumption was wrong. In general, I really liked how he took AI/ML and presented it this fresh and sexy way.

The Fall

Back in those days, I was wondering, how he manages to release one video after another, while understanding the concepts behind it, coding and doing all the necessary marketing. I assumed, that he must be some kind of genius. In terms of marketing this might be true, in terms of code he had to rely on other peoples work, basically taking stuff from GitHub and maximal putting some wrappers around it. I will come back to that later.

There were two events, which triggered his fall:

1. Students enrolled to a course of him were unhappy and demanded a refund, he handled that poorly
2. He wrote a research paper, but it actually turned out, that it was mostly plagiarised.

Course

He offered a course called "Make Money with Machine Learning" for $199. The course is teaching people how machine learning works and how they can either start a business or land a well paid job using these skills. The title itself is kind of a red flag, but I'll let that pass. He promised to take maximal 500 students, but it turns out that he actually took over 1200 students for this course. When some students were not happy with the content of the course and demanded a refund, they found out that he didn't have a refund policy. When this situation occurred, he handled it rather poorly, which he admitted himself.

You can read more about this on The Register and Reddit. So far you might think: well, this is an honest mistake, sure if you work a lot, you can forget about some details. This course was actually not his first one, back in 2017 he did an Udacity Nanodegree on Deep Learning, one of the people from Udacity, he worked with, pointed out how even back then, he didn't really give much thought for correct attribution of work.

Plagiarism

That brings me now to the part two of his fall, and this one is rather big. Here on Steem, people should be aware why plagiarism is bad and why they mustn't do it. In the world of academia it is even more strict, plagiarism is a cardinal sin and one of the worst things you can do, the only thing that compares to it is fabricating data. In both cases you usually lose your job, degree, reputation and most likely your livelihood as well. Scientific integrity is a key principle in academia. He is not in academia, but even though he often pointed out, that he dropped out of college, the things he is talking about are still closely linked to academia and therefore he should follow its principles as well.

He published a research paper on quantum machine learning, which was a very blatant copy of another paper(he changed all "we" to "I" and made up some terms, which make no sense, e.g. turning "complex Hilbert spaces" to "complicated Hilbert spaces", complex in this case referring to complex numbers), he even cited in the abstract of his paper. In the livestream, I linked above, he even brags about this paper (the livestream was released, after the first "scandal" already happened). What makes the whole thing even more ironic is, that the example he brings is about fraud detection (which btw. was also not his own work). Here is the Tweet of Andrew Webb, who first exposed this and another article from The Register on this topic.

The question on my mind was simple: Why? I think that he is smart enough, to see that people will detect the plagiarism and call him out on that. He should also know, that losing his credibility will also have negative consequences for his YouTube career (although in terms of views of his videos, it didn't really show so far) and that people will stop taking him seriously. As I said above, his doesn't seem to put a lot of thought into correctly attributing other people's work. In his videos, he gave credit to the original authors of the code, but it was rather hidden (barely mentioned in his videos) and put into a new GitHub repository, instead of forking the existing one (which would be the proper way to do it, especially if you only do incremental updates).

The aftermath

Ever since, he released two videos: one where he shows the things some project the students of his course are working on and one where he talks about tools for developers, which again draws very heavy influences from the work of others. To be fair, this video was probably finished, before the whole plagiarism scandal broke loose and was scheduled to be posted after it did, and he didn't change it. But after everything that happened, he should have been more careful with this.

He also got kicked out of the School of AI, which he founded.

Of course, it is easy to point the finger and call "Scam!", but I don't think that's the case. He was undeniably good at making videos, that educated and inspired people, but he wasn't as good when it came to putting these ideas into code. I think that he had good intentions, but couldn't really keep up with them and started taking short-cuts, which ended in him blatantly plagiarising other people's work.

You could also argue, that he himself was in the process of learning, but Vicki Boykis put it best (I can highly recommend reading the whole article):

You can learn at 50 YouTube followers. You can’t learn when you have 700k. Which is maybe where he started copying and pasting code, to keep up.

Although I have to say that copying and pasting code was something he did before he got famous. To me it seems more like he took the "fake it until you make it" mentality of Silicon Valley way too far and it came back to hunt him. The gold rush mentality around AI/ML and the pace of social media also did their part.

However, I'm curious to see what new videos he releases and if he learned a lesson from all of this. I hope for him, that he adapts a less-is-more mentality and releases fewer videos, but puts more effort into the code behind these videos.