How would you do a 'deep dive' into a particular subject or field?

July 9, 2015 1:01 PM

I'm planning on doing a few week-long 'deep dives' into AI, health/biotech, energy, and other fields to understand them better and see if I want to make them into a career path. How would you structure such a deep dive? What strategies for finding information would you use? (Some ideas inside.)

My goal is to get a sense of a field and understand it well enough to articulate the key concepts, current state of affairs, and open questions in these fields. (For AI, for example, one key concept and open question is how to build benevolent AI.)

Here are some strategies I've thought about:

• Reading a book that surveys the entire field that can act as a jumping-off point
• Talk to people (e.g. postdocs) who are extremely well-versed in the field and get their suggestions for what to look at
• Find all-around resources like Wikipedia and use those as jumping-off points
• For academic reading, focus on reading reviews which summarize findings

Any other ideas? Thanks!

posted by markbao to Education (15 answers total)

iTunesU and Coursera come to mind.
posted by hz37 at 1:04 PM on July 9, 2015

Netflix has a lot of cool single-subject documentaries, so that could be another jumping-off point for you and would get you familiar with some key names and resources that are useful in understanding the field.
posted by capricorn at 1:38 PM on July 9, 2015

Best answer: If I remember form your previous questions, you are an undergrad, so suggesting things tha tyou should be able to access on campus.

One of the things that helped me eliminate and/or continue with potential career ideas as an undergrad was volunteering in the particular setting/job if you have the time available. For example, you can intellectually explore "health" and think: doctor or physical therapist. But volunteering in a clinic can help you realize that you might not like interacting with patients, for example.

If you can identify something more specific within biotech or health, approach profs at the start of the semester and ask if you volunteer in a lab for credit. This can help you work on projects, participate in journal clubs, be part (and might even be listed as a contributor or author) of publications, etc. But this can really help explore a field intellectually vs. trying to understand it from a review article.

I think talking to postdocs is a great idea. What I would suggest for going beyond postdocs is to ask if they can recommend one of the best primary journal articles in the field/or something from their own research.

Find out if there are journal clubs and/or departmental presentations for the fields that you want to explore. These are usually for the general public, too, and you might stumble upon really interesting projects or just ways to think about questions and research, etc.
posted by Wolfster at 1:38 PM on July 9, 2015 [1 favorite]

Figure out what the most popular or highest impact journals are in each subfield of interest and scan the table of contents for issues from the past year. Read the articles that interest you to see what the scholarship in those disciplines is like.
posted by sockermom at 1:40 PM on July 9, 2015

I'd try to find a recent and up-to-date course that overviews the topic at a not too introductory level, with online materials, and go through the materials. For something like AI it would be best if there were homeworks and you did some of the homeworks (don't know much about the other fields). Be warned that I don't think a week is anywhere near enough to really accomplish your goals as stated, though you could get a start. For example, here's an MIT AI course that looks decent to me (though I don't know about its textbook, it's not AIMA).

I would probably also consider asking some field-specific questions like this on AskMe. (I see you've actually already done this for AI.) Some things I would not do: look at wikipedia except as a reference for technical material (and then cross-check it with other things), watch ted talks or the like, read a popular science book that is supposed to be surveying the field (unless highly recommended by actual practitioners, these do occasionally exist).

(For AI, for example, one key concept and open question is how to build benevolent AI.)

This is not a key concept or (interesting) open question in AI. (As you'll see from the course I linked above.)
posted by advil at 1:45 PM on July 9, 2015

this is an interesting question for me, because i recently started working part time with the aim of using my spare time to "learn more". i hadn't really thought of it in terms of doing "deep dives" like you describe, although it's an interesting idea. if it's any help, my tactics are:

• keeping track of interesting things that i might want to go back to. i do this here (sorry, self link) for things i've found on the web. but i also have a whiteboard with a whole list of possible projects.
• use a more tree-like structure, rather than separate, "linear" dives. so start working on one thing (like bruce schneier's teach yourself crypto course) and then realise i need to know more about, say, finite fields, so go off and read a book on that.
• be very relaxed about jumping around. i live in chile, so any book takes weeks to get here. so if i need a book then i'll head off somewhere else and perhaps not look at the book for months. i think it's important to have fun - it cant be a grind - or you just won't do it.

maybe that's too "me" and not very helpful. sorry. i guess one thing that might be helpful, if you're looking for where you might work, is the idea that you keep following things where they seem interesting. so if one of your dives ends when something still looks worth investigating, maybe it's worth following anyway. don't be too structured (on the other hand, some structure is necessary in your casem if you have a deadline - i'm just filling in time til i die :o)
posted by andrewcooke at 2:23 PM on July 9, 2015

Best answer: I did this all the time as a kid. Kids instinctively know how to go about it: unstructured, passion-led, playful immersion. Let your muse direct you.

Humans are famously incapable of learning new things as they get older. I think it's because they disconnect from whim-led playful immersion. I'm 52, and I seem to have missed the gene that turns off this instinct. Give me some pop-up books about, like, dinosaurs, and I'll be all over them for days. I never stopped immersing.

So I'd just dive in anywhere and devour with giddy zeal. Turn off the overarching camera of self-consciousness - the grown-up supervising self that monitors and judges - and just play.

Your knowledge will be spotty as a result; that's the only downside. But at that point, go ahead and use your grown-up supervising self to evaluate missing chunks and fill those holes in a more disciplined, directed way.
posted by Quisp Lover at 2:43 PM on July 9, 2015 [2 favorites]

If you're going to jump into the scientific literature, I recommend starting with review articles.
posted by congen at 4:05 PM on July 9, 2015

Best answer: A simple answer if you're in college is to get the syllabus from an appropriate intro class in the fields you're interested in and then read the required and suggested readings.
posted by MsMolly at 4:26 PM on July 9, 2015 [2 favorites]

The Great Courses could be a good foray for certain fields of study. Think you can download lecture series on Audible.
posted by doctordrey at 4:38 PM on July 9, 2015

(For AI, for example, one key concept and open question is how to build benevolent AI.)
Nick Bostrom's book Superintelligence: Paths, Dangers, Strategies is a good starting point.
posted by James Scott-Brown at 4:38 PM on July 9, 2015

Here's a pretty solid if not seminal introductory AI problem: Digit Recognition using the MNIST dataset. Read around on it. That's the daily grind of AI: things like that, but more complicated.
posted by oceanjesse at 8:46 PM on July 9, 2015

Response by poster: Thank you all for the great suggestions!
posted by markbao at 10:13 PM on July 9, 2015

Academic here. Look at books like Cambridge Companion to X or Oxford Handbook of X.
posted by persona au gratin at 2:44 AM on July 10, 2015

this book may be worth reading - the guy's blog is often excellent (my copy in post so can't give a review).
posted by andrewcooke at 5:14 AM on July 11, 2015