Ask HN: What would you recommend me to update myself on AI?

11 points by calebjosue 2y ago ↗ HN

Back in 2011 I enrolled in the Machine Learning class by Andrew Ng via Coursera (Didn't finish it). I also got into the AI class, same year via Coursera. By Sebastian Thrun, and Peter Norvig (Didn't finish it). Yeah, I know.

Back in 2017, I enrolled into Deep Learning Specialization by Andrew Ng, via Coursera. I finished four of five courses. I barely remember anything. [1]

In 2018 I got into Udacity's AI programming with Python, finished the course, but I didn't exercised the thing after the end of course. [2]

In 2020 I bought the book "Artificial Intelligence : A modern approach" (Fourth edition was just releases) by Norvig and Russell (Didn't read the book).

Now I am seeing all sorts of AI applications here and there, some cats dancing out of a picture I suppose. And I would like to know what's new in the field.

Yeah, your main recommendation may be to commit to finish the courses I start, I know. These days I am doing just that. But I am revisiting some elementary Math topics instead like Algebra and Geometry (Which I recently finished a book in the Gelfand series).

What's the hottest training material right now on AI? I think I will need to know some advanced Calculus prior to that, right?

I just saw some "Elements of differentiable programming" shared over here just the other day, but I suppose the thing divide itself into designing the algorithms, some other people train models using existing algorithms, and then there exists consuming existing APIs. to help you recommending something, I am the kind of person interested in "How the things works" (Even when I am not working right now, I don't have a job, and I am looking for one just in case you have something for me). But yeah. Appreciate your input.

Thanks!

[1] https://coursera.org/share/09361d90a6fc1c987af0f89502f76c9e [2] https://www.udacity.com/certificate/DJ7YNX9X

4 comments

[ 4.7 ms ] story [ 40.4 ms ] thread

p1esk 2y ago ↗

Yeah, your main recommendation may be to commit to finish the courses I start, I know.

Based on what you wrote, my recommendation is: find something you’re actually interested in, and learn that instead. AI ain’t it. And if there’s nothing you’ve interested in enough to put an effort into, that’s ok too. Have a beer, watch some Netflix, be like most people.

Hope this helps.

worddepress 2y ago ↗

I agree. I thought "I should be the kinda person who loves AI". While doing some courses was super interesting. Waiting around while the computer does a load of computation, a very slow feedback loop, and dealing with ai ops related stuff is not really that fun. Neither is the pytorch kinds of stuff (which is mostly thinking about the shape of data and what shape it will be after doing X to it). I much more enjoy app development, and AI-ifying things with API calls as needed.

calebjosue 2y ago ↗

>Waiting around while the computer does a load of computation. You know, this reminded me of Blender 3D, once you are done with your model, etc. And proceed to the rendering process, we definitively have to wait for the computer to do its thing. But you are right on what you say. Hard to come up with algorithms that not depends on massive data I suppose. Thanks both of you for your recommendations.

I am revisiting some elementary Math topics, kind of rediscovering or discovering for the first time the whole subject. But eventually I think I will open the aforementioned book (And hopefully this time I won't drop it jehehe). Having a beer, watch Netflix? That sounds cool, I am waiting for the next season of Rings of Powers.