What is the difference between artificial intelligence and machine learning?
What is the difference between artificial intelligence and machine learning?
This is a question I get asked often so I thought it might be worth writing a post about it.
If you have not been living under a rock for the past 10 years, chances are you have heard or read in the news about the words artificial intelligence (AI) and machine learning (ML). Unfortunately, articles using these buzz words interchangeably have probably confused you.
Thankfully, it is not as complex as it seems and hopefully by the end of this article you will know the difference.
AI vs ML
While AI encompasses the methods that make machines able to simulate human intelligence, machine learning is only a field of AI as shown below. Machine learning consists in building computer programs that have the ability to learn from data (thereby simulating human intelligence).
Relationship between AI, ML and deep learning
So what’s an example of something that’s in this AI circle but not in the ML one?
Most people cannot answer this question because they give too much importance to the word “intelligence” in “artificial intelligence”. However, an AI is not necessarily super smart.
Indeed, a program designed with only a series of if/then/else conditions could play tic-tac-toe and win or draw every time against Einstein. Even such a simple program falls within AI as it simulates human intelligence on that specific task.
However, it is simply given a set of rules written by a software engineer but it never learns how to play from data so it’s not a machine learning algorithm.
Thus, AI programs do not need to be intelligent, they only need to look intelligent. Despite simply following fixed rules, some AI programs can leverage computing power to surpass most humans in so called intellectual tasks such as chess.
Software engineer vs ML engineer
In general, an AI that does not exploit ML requires its software engineer to learn how to perform a given task himself and then communicate that as best as he can, breaking it down into a set of rules.
In the field of ML however, the engineer (often called machine learning engineer) does not need to learn how to perform the task himself. All he needs is to ensure he can feed task data to a program (often called a model) that relies on statistics to learn from the data how to perform the given task.
For example, last year I developed an AI that was able to spot cancer cells much better than I could, something which would not have been possible without machine learning.
PS
I hoped you liked my first post. I intend to write other AI/ML articles in the future, both technical and non technical, so if there is a specific topic in that field that you would like to better understand, please reach out to me on LinkedIn.