Machine learning/Artificial intelligence is the talk of the town. I have seen so many people trying to use machine learning for problems that can simply be solved with a simple electronic relay. Mr. Abeeb if you come across this, kindly pardon me. Machine learning is superb no doubt but what about using the right tools for the right work? This is not something we can overlook. At that, I think it’s the right time we start learning from the grassroots, knowing the basics, find a balance between sounding intelligent and gaining beneficial knowledge. Diving from Electrical/Electronic engineering into the world of computer science (Artificial Intelligence/Machine learning) is one of the toughest decisions I have made, thanks to a friend like Brother Abdul-Kareem Shamba. Eventually, I transitioned, and I have never for once regretted I did. Trust me, I am not throwing Electrical/Electronic engineering under the bus but I rather look at the possibilities of interoperability of AI/ML and Elect/Elect engineering. Moving forward, I started in this field as far back as 2020 with learning Python programming language. This took me more than three months to learn and truthfully python is a continuous process which I can say that I am still learning to date. I picked up many machine learning books, research papers, and all just to get a handful of what MACHINE LEARNING is. To me, I defined machine learning as human learning from past experiences. As humans, we are synonymous with machine learning models. Ask me how.
Ok, how?
I will tell you.
Do you agree with me that someone who grew up in a remote area all her life and known cutlass to be a knife and vice versa, this person will fail woefully when she moves out of her comfort zone. Why? Because she has learned data with the wrong label. This is what machine learning does. A good model architecture can perform woefully when trained with the wrong dataset i.e. a dataset full of noise and mislabeled. These cut across all areas of life. This brings about people with different beliefs, ideas, and know-how because we have all learned with variations of data both well-labeled datasets and wrong-labelled datasets full of noise. Anyway, I was trained with good datasets, smiles!
Oya, let us continue.
This human-AI idea helped me resonate well with AI, it gave me an in-depth knowledge of what AI and Machine learnings are. They aren't a product of magic or better put, an Angel but an imitation of humans, we have it before they do. Fine, they are not explicitly programmed but then they need us to learn just as we need our parents, peer group, and environment to learn. They can't do all things, they aren't taking your jobs we are meant to complement one another, and their importance can't be overlooked so as humans. To prove my point, I got a gig sometimes ago to build a hospital assistance model capable of translating Nupe language to English and vice versa, you won’t believe this project went down the drain. Why? Because we don’t have sufficient dataset in Nupe dialect to train our model. Here, the model is feasible but we don’t have the human needed resources to accomplish this, Mubaraq aderogba can attest to this.
We have been witnessing the geometrical progression of technology of late, guess what, technologies don't build themselves, they are built by humans and as the world progresses technology does too and the good part is they can’t build themselves they need US.