Why Artificial Intelligence (AI) is not Machine Learning (ML)
This week, I'm going to debunk one of the usual marketing tricks in our current tech society. Adding AI to any kind of software to make it new, shiny and tech-savvy. We see this term added to every slightly automated software. Many of them are using machine learning only and I wouldn't consider them as AI. So I'm going to write why I do not consider machine learning to be enough to use the term artificial intelligence.
Based on my analysis, I define Machine Learning as a field of artificial intelligence, which study statistical techniques or algorithms, which learn from data to make predictions.
Machine learning is very powerful when you have a lot of data at your disposal. For example, it is used to recognize if a cat is on a picture. One of the methods goes by serving millions of pictures to the algorithm and telling him if there is a cat on it or not. Then by sending the algorithm a picture, he tries to deduct if there is a cat on it or not, based on all the pictures and data he already has.
I find artificial intelligence to be a system of machines which show clear and human-like intelligence. To be more precise, it is a system that reproduces human behavior in such a way that you could believe this is done by a human.
The most known example is intelligent chatbots that not only understand what you write but can also adapt to your sentences to answer you valid answers with sometimes filled with emotions. They are even able to do jokes, be witty or be emotional.
Artificial intelligence is used to tackle malleable issues where the way to solve a problem needs to change on-the-fly and sometime even with making interactions with the outside world.
Example: EZYcount develops an AI which uses machine learning but not only.
At EZYcount, we build a solution and we talk about AI because, in short, EZYcount AI replicates the job of an assistant fiduciary. It collects data. Makes bookings based on that data. If needed, it checks past data of that customer to increase the certainty of the booking. If the AI is not sure how to make the booking, it communicates with the customer or his/her fiduciary to confirm the booking.
In EZYcount case, machine learning is used by the AI to look at past data and find patterns that can be used on the new data coming in. However, Machine Learning is only a little part of the complete system.
As the system is more than just then machine learning and simulate human interaction, I define it as artificial intelligence.
Be proud to use machine learning if you use machine learning. Don't pour artificial intelligence only because it sounds great!
Do you know solutions or systems that state artificial intelligence on their marketing cover but are actually using machine learning? Please do share it below.
Thank you for sharing. Have a great end of the weekend.
Vivien, CEO EZYcount