What's up guys! Today we are going to talk about AI that build AI. My name is Selim Chehimi
And this is episode 8 of AI news.
If you think that AI can't replace computer programmers in the future, you are wrong! This is the dream of
researchers: building a machine that could build other machines.
Recently, Jeff Dean, one of Google's leading engineers, announced a new project called AutoML. ML stands for Machine Learning,
if you want to know more about the differences between AI and Machine Learning,
I made an entire video about this
And I'm going to put a link to the video so that you can see it.
AutoML is a machine learning algorithm that can create other machine learning algorithm.
As you can imagine, in the long term,
This technology may replace computer programmers. If you're already familiar with machine learning,
You know how much it requires mathematics.
Did you know that 10.000 people worldwide have the knowledge and experiences needed to build these complex
mathematical algorithms? There is definitely a lack of AI talents!
Between us, if you are planning to be a machine learning engineer,
I want to emphasize that the tech giants such as google, facebook, Microsoft are paying millions of dollars a year to keep their AI talent.
This lack of AI talent is the reason why there is lag in the adoption and effective use of AI in companies.
Consequently, AI has not impacted the economy for the moment. Now, Jeff Dean's vision makes more sense and is very easy to understand:
"the work we are working on will help companies build systems with artificial intelligence,
even if they lack extensive expertise".
Artificial Neural networks are
algorithms inspired by the human brain. They can learn tasks without instructions if you give them many examples. To
have a neural network to learn a specific task,
researchers are first running 10, or even 100 of experiments to find the right algorithm.
It's hard to find because it depends on so many factors. It can depend on the size, quality, and nature of data,
, but also on how much time they have.
They also have to do many adjustments over and over until they obtain an algorithm that works well. Sometimes,
it's hard to explain for them why they did a particular adjustment. In fact, they may not know themselves! Honestly,
it's nearly considered like magic. Nowadays, those algorithms are like a black box that we don't really understand.
AutoML has been created to automate this process.
I personally believe that if something can be automated, it should be automated. The results are very promising.
For example, Google announced that AutoML built an image recognition
algorithm that is more effective than those created by humans.
Thank you so much for watching guys, if you are not yet subscribed to my channel
please do so, the goal is to reach
20.000 subscribers by the end of the year. If you want to see more of what AI can do, I linked another video for
you that is interesting.There is a playlist where you can see all the videos
I've made about Artificial Intelligence.
If you want to learn more about AI, there is a link to my blog selimchehimi.com where I'm posting articles about it.
Không có nhận xét nào:
Đăng nhận xét