This article is in Portuguese here
After a few months learning about Machine Learning and participating in the Behind The Code development marathon offered by IBM here in Brazil, in the course of the challenges I was amazed at how big companies are creating AI solutions that are so simple to implement that we do not need them any more. a line of code for it!
Taking AI from gods and giving to humans
"He was a defender of humanity, known for his clever intelligence, responsible for stealing the fire of Hestia and giving it to mortals."
Wikipedia Brazil about the Greek Mythology of Prometheus
Python, R, Julia, hours and hours of reading posts, books, articles, watching the most diverse videos and watching classes until I understand a little of this incredible universe that is AI. But now and then I wondered if it would be possible for AI to become something simpler, because it now seemed that it was necessary to be an extremely skilled professional, almost a "God" of mathematics and programming, so that one could work with AI, but it seems that this is already changing and like the Promethean legend, the AI is coming out of "Olympus", and being delivered to "mortals."
While True: Learn was the first idea that showed me how AI could be simpler, the game teaches you through streams like creating classifiers. In it, you are a freelance programmer who performs the most varied work of ML.
The works are very similar to real-world problems, in addition to the game provide content to learn how it works in real-life game models, if you want to see more about the game it is for sale on Steam here.
Life imitates art
IBM Watson Studio with Machine Learning
IBM with Watson, Microsoft with Azure ML Studio, Jupyter with Jupyter Labs, several platforms are creating ways to make AI no longer a complex thing to do.
Instead of hours writing lines of code to assemble a predictive model, use flowcharts to assemble your model. When working with Watson Studio, and realizing that without a line of code I could create an excellent classifier of image recognition was something amazing.
With this, you realize that now, AI is now for all . Over time, more and more areas will benefit from this, these platforms are taking away the need to have a highly skilled Data Science professional and providing resources for practitioners from other areas such as medicine, economics, biology to be able to use AI to their pleasure . And that breakthrough was in a matter of a few years, which surprised me most in this process.
Watson Creating as classes for training from images in .zip files recognizing each type of dog.
The challenge of AI is not so simple either
"Making simple the complicated is easy, making the complicated in simple, this is creativity"Charles mingus
Of course, those who possess a theoretical ML bag can create better predictive models as needed. We can not forget that these platforms are abstracting complications so that people who are not IT area can take advantage of AI, but they are platforms of support.
Asking the right questions about what your data can show is a challenge that will still require of our heads, in being creative and innovative so that we can extract intelligence from the data that we have, in the end, the work is still ours.
But now the competition is changing, now it's no longer the best Data Scientist or ML Engineer, but whoever collects, organizes the best data and asks the best questions on these data. Now the important thing is guidelines and data sources that will be used.
But seeing all this, I see how Data Science has evolved in such a short time, the future looks bright, and it ends up making me very excited, as they always say, technology has come to stay haha.
Sources:
https://luden.io/wtl/
https://www.ibm.com/watson/br-pt/
https://en.wikipedia.org/wiki/Prometheus
Top comments (0)