Reddit seems to be a haven for everything you can imagine on the internet. A lot of things that often go overlooked. However, there is a post about the so-called AI guru named Siraj Raval exploiting students with charged them $199 for machine learning course that copy-pasted other people’s GitHub code and banning students from asking for a refund. There are a few lessons that we can learn from this story so we are aware of this problem and make sure it won’t happen on us in the future especially to naive and AI practitioners that self claim as ‘guru’.
Lesson number 1: If you can make money, why bother charge people when teaching them how to make money?
Why bother to charge people for a machine learning course if you can make a lot of money using machine learning? It doesn’t add up, right? Let say if you already made a million-dollar but at the same time, you willing to teach people how to make money by charging them for a thousand dollars? At that time, you already too busy making money, more and more, no much time to focus on other things. Nowadays, it happened to many people so-called gurus that offer an expensive course to make a ton of money even though they don’t have any proven business.
Lesson number 2: Hard-won experience doesn’t mean everything you said is wisdom.
Take an example from the AI debate between Elon Musk and Jack Ma, Elon Musk looks unhinged and realized Jack Ma didn’t have any idea about artificial intelligence. But Jack Ma’s wealth shows that any of his words are wisdom. The same goes for Siraj Raval, he successfully to be a social media influencer on youtube with 700k subscribers. With his influence, he successfully partnership with Udacity about his machine learning course. But it didn’t go well. According to Ray Phan:
“As an example, he designed an assignment where you had to use linear regression to predict stock prices,” said Phan. “Linear regression in the sense that it’s a first-order predictor where you predict stocks with a straight-line approximator. This is utter garbage and not representative of what actually happens in the industry.
Lesson number 3: Naive people easy to be exploited with low tier educational content that is mostly inflated with hype.
AI can take our jobs. AI can destroy the world. AI can think like a human. You must stop believing all these myths because its unlikely to occur in the near future because companies look at AI technology as a way to increase the human workforce and allow them to operate in newer and smarter ways. Also, creative capability, emotional intelligence (such as empathy) and strategic thinking are still nowhere near. Lack of experience, wisdom, and judgment can be exploited easily by snake oil salesmen with plenty of motive for spreading fear and distrust to make fast money.
Lesson number 4: Snake oil salesman.
In business especially for a startup, never ever deceives customers in order to get money from them or gain fame. It’s really not a good best practice. Why? Distrust can destroy your business when it comes to a big pile of money. Siraj knowing his video and course can’t give so much knowledge since it’s only at the basic fundamental, unaccurate and didn’t guarantee that the trainees can get jobs elsewhere especially in San Fransisco bay. He is not only a snake oil salesman but damaging the whole data science field as well.
Lesson number 5: Don’t be a jack of all trade. You will be a master of none.
In Siraj’s Youtube video, there are a lot of things covered. From linear regression, neural network, deep learning, virtual reality, blockchain, cryptocurrency to quantum computing. Even he teaches how to make a financial freedom machine using machine learning. Recently he just wrote a new paper, The Neural Qubit. If you take a look at his book on Amazon, the reviews explain all about his knowledge about blockchain. I’m not sure which area he will be focusing on after this.
Lesson number 6: Machine Learning is not as easy as 1,2,3.
If you think machine learning is easy to learn, you are completely wrong. Now we are talking about specifically machine learning, not AI as a general. We are not talking about neuroscience, robotic, expert system or self-driving car in this section. Programming is easy to learn, compare to machine learning, it’s too complex compare to basic assignments that you found on the internet such as Kaggle or in the university, you need to have a basic understanding of data science that requires math, programming, and specific business domain.
Besides that, a data scientist also needs to know how to understand data, how to process the data, conventional machine learning techniques such as classification (Logistic Regression, Naive Bayes Classifier, K-Nearest Neighbor, Support Vector Machines, Decision Trees, Random Forest or Neural Networks) and clustering (k-means). Then we can proceed with deep learning and so on. In the real world, the main challenge is to gain the highest accuracy and collecting the dataset to improve the result from time to time. You can’t simply create a dummy data or grab a few datasets from public repositories to train your data. The data cleaning process takes an average of 60% of the time.
Lesson number 7: If you can’t explain it simply, you don’t understand it well enough.
Be an educator like me it’s not an easy job. Even you are an expert, you still need to revise your explanation into simple words that layman and naive people can be understood. If you can’t make them understand, the main objective of the training will simply fail. As a result, your rating will be dropped and people will start to lost trust. The same goes for Siraj’s videos, the terms used are quite confusing, shoddy and clearly didn’t really know what he was talking about.
The conclusion is to stay hungry and stay foolish. Keep learning every day and do some due diligence first before entering or accepting something as your final judgment and wisdom.