“Don’t start with definitions and theory. Instead, start by associating the topic with the results you want. “
– Jason Brownlee, PhD, Machine learning management
If you’ve read some of my articles, you may have noticed I love theoretical artificial intelligence. It has a philosophical touch that amazes me. I want to think about the future of the industry and what it means for us. How does AGI shape our lives? Do we even create it unnoticed? Is it dangerous or calm?
I also want to read about historical grounds. How were cognitive sciences and computer science separated at birth in the mid-20th century? How did symbolic artificial intelligence fall from grace to allow neural networks to step in and eventually reach the hegemonic state they enjoy today? Why has artificial intelligence suffered from the winter season?
I also want to know where things are coming from. What is convolution? How does a transformer, based solely on attention, overthrow other paradigms in recent years? How is it possible that adding parameters makes models qualitatively more efficient?
All of these questions appeal to me like a moth to a light bulb. Still, when I started learning artificial intelligence and took the courses of Andrew Ng and Geoff Hinton back in 2017, I knew that the philosophical, historical, and theoretical aspects of artificial intelligence wouldn’t get me a job. I wanted to work on a real life project. I wanted to get my hands dirty. I wanted to gain the know-how of actually building something from scratch. Books couldn’t teach me that. Stories and in-depth reflections couldn’t teach me that. I needed practical knowledge and useful skills to keep the project from falling apart.
But the path is not the same for everyone. In theory, starting would make sense in two situations. First, if you’re studying computer science at university, you’re likely to learn the basics of artificial intelligence, even if you don’t want to. But since you’re still in college – and the potential success of the project doesn’t rest on your shoulders – it’s probably worth taking advantage of it. Theory and history are not the most urgent features of artificial intelligence, but they are certainly an advantage when competing with other people.
Another situation is if you are going to college. Scientists and researchers know the pillars of their research. They simply do not use artificial intelligence to build a company or project; they create those algorithms. They know why transformers are more successful than other models. They know why symbolic artificial intelligence didn’t work. They know because they created it. And they also created in-depth learning. Every major breakthrough in artificial intelligence came from scientists who knew the theory behind practice.