You Should(n't) Learn Machine Learning

ML

Last Edited: 7/5/2024

I am writing this blog to reflect on my experiences with machine learning so far.

ML

Machine learning (ML) has been a hot topic in recent years as we observe the impressive performances of ChatGPT, Dall-E, Stable Diffusion, Tesla, and more. But before diving into the endless world of ML, let's step back for a second and ask, "Is it really worth learning ML?"

ML as a Career

If you think you can become an ML engineer and earn a substantial salary, you are not wrong. According to Indeed, a machine learning engineer in the US generally earns between USD 100K to 250K. Additionally, the US Bureau of Labor Statistics predicts a 21% growth in jobs between 2021 and 2031, indicating an increase in job postings. But should money be your primary motivation? Absolutely NOT.

ML is HARD

Machine learning is an interdisciplinary field that involves many areas of math, science, and engineering. You have to learn a multitude of concepts to become remotely useful. Because of this, almost all job postings for ML-related positions require at least a master's degree, a PhD, or even a publication of a novel paper in a well-established journal. You will have to sacrifice a lot to become an ML engineer, which justifies the high salary.

ML is COMPETITIVE

In addition to being challenging, ML is an extremely competitive field. The smartest people on this planet are competing to develop the best machine learning models, releasing new, high-performing models every single day. Universities and companies are boosting their expenditure on computing resources to accelerate development, making it almost impossible for any individual developer to keep up with the pace and compete in this field. Even if you become an ML engineer, you will need to constantly strive to keep up with the latest advancements to maintain your profession.

ML is TEDIOUS

Many newcomers see the shiny examples of ML applications and dream of becoming engineers who solve all the problems with the latest technologies like magic. But the harsh reality is, ML is not magic or a one-size-fits-all solution. It is just a powerful tool that is insanely hard and tedious to master, requiring many monotonous tasks to make it work. You will have to spend days and hours cleaning up data on spreadsheets, debugging thousands of lines of code (often cryptic), and monitoring training processes. Even behind the shiny technologies we see in the news lies a tremendous amount of tedious work done by engineers.

Who should learn ML?

In my opinion, you are best suited to pursue ML if you meet the following criteria:

  • You LOVE math and science
  • You LOVE building things
  • You LOVE doing tedious work for others (even unnoticed)
  • (Optional) You have a mission in life that involves ML

The last point could apply to someone in fields like healthcare, finance, or transportation, who is ready to go to great lengths to innovate. If you are only looking for money and fame, I must tell you that ML is not the right field for you. However, you need to try first to see if you like something. I highly recommend you at least give it a try to see if you meet the above criteria. If you decide to start learning ML or are currently learning ML, I wish you the best.

Resources

Back to Top↑