AutoML, the future of machine learning?

Machine Learning is getting more and more important in our everyday life. Nearly every company nowadays wants to work with machine learning and big data, but one problem is that there are more machine learning projects, that the companies want to start than there are machine learning engineers that are able to work on these projects. People who are not familiar with machine learning are usually unable to work on these projects. In the end many machine learning projects do not get started or start much later than the companies want them to start. Additionally machine learning engineers can make a very high salary demand, due to the fact, that they just choose another project, if one company does not want to pay them the amount they desire.

AutoML solves these problems in a very elegant way. Usually a data scientist will analyze the problem and the data regarding your machine learning problem and to then build a model and train that model to solve your problem. The machine learning engineer needs to have certain knowledge and experience, to choose a good model and suitable data preprocessing steps to guarantee good results. AutoML solves these problems, a machine learning engineer would usually solve, in an algorithmic way.  Hence an AutoML algorithm produces you a good machine learning model depending on the data you have and the machine learning engineer is not needed that badly anymore.

This technology has the potential to change the way we experience machine learning in everyday life completely. Nowadays you have to study machine learning to solve machine learning problems, if you run a company and you got a problem that you can solve with machine learning you need to hire a machine learning engineer and getting in touch with machine learning has certain barriers. With AutoML everyone is able to use machine learning and everyone is able to solve ones machine learning problems in a good way. Thus this technology has the potential to spread machine learning among many people. 

This may has positive and may has negative consequences. One the one hand AutoML has the potential to make companies less depended on machine learning engineers and let them solve their machine learning problem on their own. Furthermore this technology makes it more easy to get in touch with machine learning, because a machine learning newbie would be able to achieve good results even with little knowledge. On the other hand machine learning is a dangerous technology, Many people warn of the consequences the increasing performance of machine learning algorithms may has. Elon Musk for example said: "A.I. is far more dangerous than nukes". The idea that a technology exists, that is more dangerous then nuclear weapons and to which everyone has access. can be somewhat frightening. 

What do you think about AutoML. Is this a technology that you see as positive or negative? If you want to learn more about AutoML please let me know in the comments, in this case I would go more in detail in future posts. If I got you excited and you can not wait to learn more about AutoML, I suggest you to read https://dl.acm.org/citation.cfm?id=2487629 (Unfortunately you need access to this paper, but if you are registered at a university, you should have access)

Nino