I have the impression the content is a bit outdated here.
It’s clear that e.g. ChatGPT - on of several recent (very) deep language learning networks - is perfectly capable of writing emails in response to a customer’s request. Not (always) without errors, but far beyond what is considered “feasible” in this course …
2 Likes
You’re right, but the problem is that the State of the Art in AI (SOTA) changes on an almost daily basis these days and this course was published a couple of years ago. Here’s another thread about this exact question.
3 Likes
I’d like to add a bit to @paulinpaloalto post:
Although the content you’ll find may seem outdated, in reality it is not, in my opinion. Current SOTA, for instance ChatGPT, is based on the 2017 paper Attention Is All You Need. The ChatGPT is basically that model of attention, with some improvements, but in essence it is that. Same with YOLO 8, and so many new models. All of them are based on concepts that come from a few years ago.
If you are interested in current models, I think that you have to start with the basics, which are basically the content included in the specializations from DeepLearning.ai. If someone tries to understand an Attention model (ChatGPT) without understanding neural networks, forward/backward propagation, flattening layers, dropout layers, other regularization technics, embeddings, etc, that person will have a hard time.
My conclusion and recommendation when I am asked is: Start with the MLS, follow with the DLS, and then pick one of the other specializations. Spend some weeks or months learning all this, which are the foundation, and then move on to current matters.
I recommend it because this is the path I followed. I don’t know if I could approach the current models that I am using without having had the foundations I learned in these specializations.
You will not be wasting your time 
Hope this helps.
Juan
5 Likes
Thx for the references !
1 Like
Hi Juan - Do you recommend the Deep Learning Specialization here as a next step after AI4E?
Hi @Rachel_Theriot ,
If you feel comfortable already with the concepts of linear and logistic regression, neural networks, decision trees, forward/backward propagation, loss and cost function, gradient decent, then I would say yes, jump into the DLS where you will learn all this and more in much more depth. You’ll learn to actually implement these and more in full detail. However if you feel like you need to get a better grasp at the general concepts, then I would recommend the Machine Learning Specialization first, and then the Deep Learning Specialization.
2 Likes
The other key point to note is that MLS and DLS both require you to have reasonable competence as a python programmer before you start, whereas AI4E does not require any programming. If you are already an experienced programmer in other languages like Java, JavaScript, C# or C++, then you can probably just read a few online tutorials about python and then pick it up as you go through the courses. But if this would be literally your first exposure to any form of programming, then you really should consider taking an “intro to python” course first before taking MLS or DLS. There are lots of good python courses available on Coursera.
1 Like
Here a recent thread on coursera courses covering python programming: Require Guidance - #4 by Christian_Simonis
Have a good one!
Best regards
Christian