In 1950, when Isaac Asimov wrote “I, Robot,” he introduced us to the “Three Laws of Robotics,” which are three rules that govern the behavior of robots. Although we are still years away from creating genuine artificial intelligence, one of our most significant achievements in that endeavor is the development of machine learning, which helps robots shape their behavior to coexist with humans. Similarly, AI models such as Midjourney and ChatGPT operate based on machine learning.
Machine learning is part of the Artificial Intelligence field that focuses on designing algorithms and statistical models that allow computers to learn from data and make predictions without explicit programming.
Arthur Samuel coined the term “machine learning” in 1959. His work helped develop fundamental aspects of modern machine learning, and thanks to him, artificial intelligence has come a long way.
Types of machine learning
There are various types and classifications of machine learning, for example:
- Supervised Learning
- Unsupervised Learning
- Reinforcement Learning
Supervised Machine Learning
In Isaac Asimov’s fictional universe, the robots use a type of machine learning called Supervised Learning. They were designed with the Three Laws of Robotics, and all their actions and outputs must comply with these laws. This method of learning is known as Supervised Learning.
The main objective of Supervised Learning is to learn from a labeled dataset, where each input data point is associated with a corresponding output. By comparing inputs and outputs, the algorithm can better understand the data and make more accurate predictions when new information is given.
Common examples of Supervised Learning are:
- Email Spam Filtering. When we receive well-known spam emails, they are automatically filtered from our inboxes based on content and metadata.
- Image Classification. When the algorithm recognized images based on previous information and classified them in the image of “busses” or “trains.”
Unsupervised Machine Learning
On the contrary, in unsupervised learning, no data is given. The algorithm reads the data and identifies patterns. Based on that information, it creates new data.
Common examples of Unsupervised Learning are:
- Clustering. Group a diverse set of data based on similarity: app usage or purchasing behavior.
- Generative Models. These algorithms can replicate data and create new ones that resemble it. Here, the “Art Models” like Dalli-E or Midjourney could be an example.
Reinforcement learning
It is very similar to Supervised Learning, but the System learns by doing things here, and the environment provides feedback through rewards and penalties. Those external rules become the rules of the machine, shaping its behavior.
Maybe this is the most known example of Machine Learning (or it was before generative learning models). A typical example of this kind of learning are:
- Autonomous Vehicles. TESLA automobiles are autonomous vehicles that learn to self-driving by doing it and learning about the environment, such as other driving styles, traffic signs, etc.
- Intelligent Recommendations. Probably you’ve seen this. When surfing through a website, mainly an e-commerce site, you start receiving recommendations for articles based on what you have already seen. They also created some newsletters and special coupons for you.
Those are the most common and usually known kinds of learning models. There is a lot of information about machine learning, but as you can see, it has been introduced previously. Artificial intelligence has been around for a long time but is trending now.
So don’t worry, machines will not replace your job… yet. But you should learn to use it, like any other tool. You can check out our article about AI Tools as a Complement to My Day-to-Day Job.
But if you are looking for a job, take a look at Aviada Jobs.
PS. The cover of this article was made with ChatGPT 4.0
