Hey!
I have been learning machine learning through the machine learning specialisation course and am about to complete supervised machine learning.
But I noticed that the machine processes data provided by us, based on the algorithm devised by us. How is AI involved in this? Shouldn’t we learn how to make AI predict the algorithms, rather than creating algorithms ourselves?
@AdityaK Well, I think what you are speaking of would be ‘unsupervised learning’ + ‘reinforcement learning’.
Things do move quite fast in this field, but I am not sure we are there yet.
Personally, I am also of the mind… Would this be such a great idea ?
I do sternly feel we ought to consider what we are making, before we ‘make it’. Or there is a responsibility implied.
Hey,
So I dug a little deeper and it turns out that the AI technology hasn’t actually reached the level of generating its own algorithms.
However, it also came to my attention that an emerging field called automated machine learning(AutoML) specifically aims to do that and automate most parts of ML. It does seem to be a promising and interesting subject and I would love to learn that…
But anyways, thanks for your insights.
@AdityaK Well, yes there is ‘AutoML’, but I feel this is not what you were talking about-- That will try out a bunch of established models on your data and find the ‘best one’
I.e. Highest accuracy.
My only personal problem with this is you don’t need to know anything about how the models actually work to implement them-- So when something ‘goes wrong’, you will have no idea why.
I greatly encourage you to keep studying, but also just remind we have a responsibility here for what we create.