Avoid These 5 Common Mistakes Every Novice in AI Makes

Image By Author

Have you heard the following saying by Albert Einstein?

Insanity is doing the same thing over and over again and expecting different results.

It is a perfect reminder for those starting their AI journey. As a beginner, it's easy to feel overwhelmed by the vast amount of information and resources available. You may find yourself making the same mistakes that countless others have made before you. But why waste time and energy repeating those errors when you can learn from their experiences?

As someone who has spoken with experienced practitioners in the field, I've always been curious to learn about their AI journey. I quickly discovered that many of them encountered similar challenges and pitfalls early on. That's why I'm writing this article—to share the 5 most common mistakes that novices in AI often make, so you can avoid them.

So, let's get started:

1. Overlooking the Fundamentals

As an AI beginner, it's easy to get excited about flashy algorithms and powerful frameworks. However, just like a tree needs strong roots to grow, your understanding of AI needs a solid foundation. Ignoring the math behind these building blocks can hold you back. Frameworks are there to help the computer perform calculations, but it's important to learn the underlying concepts instead of just relying on black-box libraries and frameworks. Many beginners start with tools like scikit-learn, and while they may get results, they often struggle to analyze performance or explain their findings. This usually happens because they skip the theory. To become a successful AI developer, it's essential to learn these core concepts.

Determining what skill sets separate a good AI developer from a novice isn't a simple, one-size-fits-all answer. It's a mix of several factors. However, for the purpose of this discussion on fundamentals, it's important to emphasize the significance of problem-solving, data structures, and algorithms. Most ML companies will assess these skills during the recruitment process, and mastering them will make you a stronger candidate.

2. The Jack-of-All-Trades Fallacy

You might have seen profiles on LinkedIn claiming expertise in AI, ML, DL, CV, NLP, and more. It's like a long list of skills that can make your head spin. Maybe it's because of social media or the trend of being a "Full Stack Developer" that people compare AI to. But let's be real here, living in a fantasy world won't help. AI is a very vast field. It's unrealistic to know everything, and trying to do so can lead to frustration and burnout. Think of it this way: it's like trying to eat an entire pizza in one bite – not exactly practical, is it? Instead, focus on becoming really good at one specific area. By narrowing your focus and dedicating your time to mastering one part of AI, you'll be able to make a meaningful impact and stand out in the competitive AI world. So, let's avoid spreading ourselves too thin, and let's concentrate on becoming experts in one thing at a time.

3. Stuck in Tutorial Trap

I think the biggest mistake beginners often make is getting overwhelmed by the countless online tutorials, courses, books, and articles available when learning AI. Learning and engaging in these courses is not a negative thing. However, my concern is that they may not find the right balance between theory and practice. Spending too much time on tutorials without actually applying what they've learned can lead to a frustrating situation known as "tutorial hell." To avoid this, it's important to put your knowledge to the test by working on real-world projects, trying out different datasets, and continuously working to improve your results. Additionally, you'll notice that some concepts taught in courses may not always work best for specific datasets or problems. For instance, I recently watched a session on Aligning LLMs with Direct Preference Optimization by DeepLearning.ai, where research scientist ED Beeching from Huggingface mentioned that although the original Direct Preference Optimization paper used RMSProp as an optimizer, they found Adam to be more effective in their experiments. You can only learn these things by getting hands-on experience and diving into practical work.

4. Quantity Over Quality Projects

When beginners want to showcase their AI skills, they often feel tempted to create numerous projects to demonstrate their expertise. However, it's crucial to prioritize quality over quantity. I've observed that people working in big tech companies often have 2-3 strong projects on their resumes, instead of 6-10 small or mediocre ones that many others include. This approach is not only beneficial for job prospects but also for learning. You can get a better understanding of the subject matter. Instead of following YouTube tutorials or building a bunch of average projects, consider investing a month or so of your time and energy into projects that will have long-term value. This approach will steepen your learning curve and truly highlight your understanding. It can also make your resume stand out from everyone else. Even after securing a job, you won't struggle much when transitioning to the actual work.

5. The Lone Wolf Syndrome

I understand that different people have different work preferences. Some may prefer working alone, while others seek support. For beginners in machine learning, it can be overwhelming, and working in isolation may hinder your growth. I highly recommend engaging with AI communities on platforms like Reddit, Discord, Slack, LinkedIn, and Facebook. If you're not comfortable with communities, consider finding an AI mentor for guidance and support. Discuss your projects with them, seek their advice, and learn about better approaches. This not only makes the learning process enjoyable but also saves time. Although I don't encourage you to immediately post questions or reach out to your mentor as soon as you encounter a problem, you should always try to solve it yourself first. But after a certain point, it's okay to seek help. This approach saves you from burnout, enhances your learning, and in the end, you'll feel good about yourself for trying and gaining knowledge about what didn't work.

50-Day Challenge: Dare to Accept and Level Up Your AI Skills

Throughout this article, we've discussed the 5 most common mistakes that beginners should avoid at all costs.

I have an EXCITING CHALLENGE for all of you. As a responsible member of this community, I want to encourage you to take action and apply these tips to your own AI journey. Here's the "50-Day Challenge":

1. Write "Challenge Accepted" in the comments section below. (Reload the page if you cannot see the comment section - it may take some time to appear.)
2. Spend the next 50 days focusing on these 5 tips and implementing them in your AI learning.
3. After 50 days, return to this article and share your experiences in the comments. Tell us what changes these tips brought into your life and how they helped you grow as an AI practitioner.

I'm eager to hear your stories and learn about your progress. Additionally, if you have any suggestions or additional tips for fellow readers, please share them! Let's help each other grow.

Kanwal Mehreen Kanwal is a machine learning engineer and a technical writer with a profound passion for data science and the intersection of AI with medicine. She co-authored the ebook "Maximizing Productivity with ChatGPT". As a Google Generation Scholar 2022 for APAC, she champions diversity and academic excellence. She's also recognized as a Teradata Diversity in Tech Scholar, Mitacs Globalink Research Scholar, and Harvard WeCode Scholar. Kanwal is an ardent advocate for change, having founded FEMCodes to empower women in STEM fields.