How to start learning about AI? Planning to take a year off work (2018) to learn


#1

I’m planning on taking next year off work and dedicating myself completely (70 hours/week) to learning about computer science and programming, with a focus on AI.

I currently have 0 experience in any of this.

I’m aware that I’m unlikely to reach a high level in the year but hopefully it will provide a solid foundation.

My question,

If you had to go back and learn everything you know now from scratch, how would you structure your time over a 1 year period?

Any other tips for the project?

Thanks,

Jonathon


#2

got no experience with hands-on Ai programming, but if I could take 1 year off to learn CS and coding all over again, focusing on ai, I’ll assume money isn’t a big issue, given the initial variables and so I’d certainly go for any highly prestigious institute such as mit, make the best out of that year and quit.

regardless of what kind of help you can get, though, the single most important aspect to properly consolidate any learning is self-interest. time invested is only second, at best. so, instead of focusing on the technology or stack, focus on what you want out of it. what you want to achieve.

and, for 1 year, make it a very small goal, something you think you could accomplish in 3 months. if you do it that fast, great, you’ve got 9 more months to either polish it up or do new things. if not, well, you still have enough time to run for it until it’s done.


#3

And if… money is an issue? :smiley:

Focus on online courses? Get into a big-data company?

What’s your take?


#4

funny you ask. I’m giving that advice while following it up myself. I can’t afford mit’s declared tuition. heck, I hardly know anything about getting to mit. still, since april I’ve been making up my mind in favor of finding such a place, with a good chance to being in mit itself. in my case because, from experience and observation, that’s probably what will bring me best results in terms of better funding my own life as a fundamental milestone for achieving my personal goals.

now, there are many other assumptions I took about the question, which may seem obvious which means it could also easily go unnoticed. ai is a very narrow study field considering all study fields together and yet, if you want to learn how to code ai you’ll have to narrow it down much further as the amount of data to digest can fill a lifetime of research. so if you’re really challenging yourself to anything in this direction and able to “take a year off any job”, I also assumed you’ve got a certain “adventurous” mindset ready to take risks…

so, right now I see two takes on this:

  • the one take you expect from your question, “okay, there’s no money”.

you could rationalize this in a few ways. maybe assume there’s simply no way to turn such a big issue into a small one or be in practice unable to find a way out, like the majority of people including myself.

in case you do give up here, while still being able to not get depressed and focused on your personal pet project drive (which in itself is a big achievement) you could simply keep trying to build it by any means necessary. get to any free course. get out of it when it stops making sense. follow the paths and move away from them if you feel like it. try as much as possible to feel right on what you’re doing. and keep track of progress.

eventually you may build some kind of rudimentary ai yourself, independently. even if not, you’ll probably learn a lot more than trying to blindly follow any pre-arranged steps, as “coding in ai” is still a field too new to have any consolidated path to follow and, at the same time, too complex to even consider building such a path to fit anyone, as a course is supposed to be.

  • the one take you probably didn’t expect, “fuck money”

as someone who made less than U$ 10k per year over my 35 lifetime,
I actually said “big issue” in my previous post on purpose. because money is always at least a small one.

of course I had to always save as much as possible. I never bought a car, for instance. and I may have a whole lot more of health issues than I would’ve otherwise. but I also never felt the need for much more money.

intermission - okay… as I’m trying to think on how to better explain this idea, I also keep thinking about how little time I’ve got right now and how I’m going too far! let me try to shorten it up and I’ll come back later to edit/continue if needed.

in one hand, it may be better to spend this year or the following ones on trying to get into mit (or any such place) until you do than getting any kind of practical experience with ai (or whatever). I’ve got plenty of practical experience with programming to prove me that it means too little as a “sales card” if compared with any experience getting into such a prestigious group. and different from trying to build a meaningful project, simply being accepted by a nice group means a lot more in society than finishing up a project that’s only meaningful to you or that will get outdated in 5 years. it’s eerly alike school 101, in a way.

in the other hand, there’s a chance you could have enough of what it takes to get on school debt (or some kind of debt). despite of what most financial healthy people might say, getting into debt can be a good thing. if it’s a debt towards building something you truly love and with people whom you and most of the world can trust, this means someone is willing to invest in you for proving you can. quite reassuring of following a path. almost nobody is born with the luck of not needing to pay this back, but well, that’s just life. nothing is fair in it.

now, of course, all that’s just my 2 cents. and thanks for asking! :wink:


#5

I’d start with the A16Z AI Playbook http://aiplaybook.a16z.com


#6

Not sure how useful this is, but you might be able to glean some habits from this post.
https://www.scotthyoung.com/blog/myprojects/mit-challenge-2/

Ultimately the guy decided to try to complete the entire MIT cirriculum for CS in 1 year, which you probably won’t need quite that, but it wouldn’t be bad to have a good chunk of that basis.