What do you mean you love AI?

“I love AI. I want to be a computer vision engineer. I want to write vision algorithms. I think I will be a perfect fit in your company.”

With hundreds of hours of recruiting experience, I have heard this sentence at least dozens of times. There are even more such expressions in the application letter sent by the candidates. And my tiredness is piling up.

To be clear, I don’t mean this kind of passion is not good; what I mean is that 80% of the candidates know almost nothing about AI. It is even worse. Being passionate about a wrong subject.

Let’s have a look at the reasons.

#1. Lack of practical experience.

What does practical experience mean?

Research in a company doesn’t count.

Coursera/Udemy/Udacity projects don’t count.

University projects, of course, don’t count.

Practical experience means you do it with the pressure of deploying to the end-user. You have to care about reaching a 99.9% accuracy rate. You have to care about insufficient and imbalanced datasets. You have to care about the deployment process on a non-Linux machine. You have to care about the millisecond level of inference speed requirement. You have to care about all edge cases in a dynamic, turbulent end-user environment. You have to care about the safety concerns of the end-user about the usage of artificial intelligence.

In what we call side projects, the dataset is prepared, cleaned, and balanced. You only need to copy-paste existing deep learning architecture, set several hyper-parameters, and wait for the confusion matrix.

This is not called practical experience.

#2. Fear-Of-Missing-Out

AI is one of the industry's most used buzzwords to get headless investors and funding. Similarly, it is also the simplest way to explain what you are working on to an outsider. But by no means should it occur in a technical interview. Artificial Intelligence in 2023, as a collection of knowledge fields, is far beyond a single person’s comprehension capability. Yet, after being bombarded by the press about ChatGPT, Midjourney, and trying a few built-in models from PyTorch or Tensorflow, someone claims he/she loves AI and wants to work in AI.

If you are like that, do yourself a favor and watch a few YouTube videos or House of Cards; you will also love making videos or playing political mediations.

#3. Solving the wrong problem

When people say that they love AI and they want to work with AI, it feels very confusing. AI is a tool. It's not the aim. It is like a runner claims he enjoys running because he likes Nike; an architect says he enjoys building houses because he loves AutoCAD; or a doctor says he enjoys saving people’s lives because he loves syringes. It completely misses the point.

Obsession of a tool will ultimately lead to non-optimal design. For someone who has a hammer, everything becomes a nail. When you're so focused on using some deep learning architectures, you're missing a big range of opportunities to solve a problem. This seems trivial. We can brute force to solve the problem with any viable tools in all cases. However, selecting tools is a significant decision for a company whose essential business goal is to increase revenue. Technical debts, extended developing time, and unnecessary development environment investments will saturate the company’s focus and customer traction.

With all the news about AI tools, how great they are, and how much we can improve our lives, it seems that we are already on the highway of reaching artificial intelligence. Technology is advancing so fast, and it makes people fear. We fear that we will be left behind. We fear that we are not able to contribute anything to the world. We fear that we are analyzed through and through by algorithms. We fear that robots will replace us.

However, our fear shouldn't turn into a reckless career path selection. Be open-minded and learn about what is possible and impossible. Find your ultimate goal and then navigate with a set of skills. You will find what you love.