There's two different (but related) questions here, so I'll try to give my thoughts vis-a-vis each. What do you think about topics in AI other than ML/DL? Bro

There's two different (but related) questions here, so I'll try to give my thoughts vis-a-vis each. What do you think about topics in AI other than ML/DL? Broadly speaking, I think there is a lot of reason to study/research areas other than DL. That's not intended as a slight against DL research... it's rather just to say that A. I think there is a lot of "meat on the bone" with regards to some other approaches, and B. more specifically I think a good path towards real progress on "smarter" AI's is integration of different techniques, including - but not limited to - Deep Learning. The rationale for that belief is roughly something like this: ANN's may in principle be capable of achieving everything up to AGI. Artificial neurons are only loosely modeled on real (biological) neurons, but nonetheless, the brain is a neural network in some sense and one might speculate that if we made our ANN's deep enough and wide enough, that any and every kind of intelligence could emerge. Maybe. However, this strikes me as basically trying to build an artificial brain in a way that is much less efficient based on what we know so far (in terms of intelligence/energy_consumption at least). But I see use of some other techniques as a way to "shortcut" implementing some kinds of reasoning (let's speculate the use of backward/forward chaining, resolution, etc. for logical inference just to use an obvious example) and using them where they are strong, and using, say, Deep Learning where it is strong. Other "things" that might fit into this model include (being very speculative here) pretty much any "old school" or less trendy AI approach that isn't DL - maybe Algorithmic Probability / Solomonoff Induction, Algorithmic Game Theory, Bayesian Networks, Inductive Rule Learning, Genetic Programming, Metaheuristics of various sorts, Abductive Reasoning, Automated Planning, use of "cognitive architectures", etc., etc., yadda yadda. So would I spend time on non-DL AI approaches (not to exclusion though)? Yes. Yes I would. And I do. In fact, just recently I've been going back and doing a deep dive back into a lot of the "old skool" AI approaches for logical inference from the 80's, alongside the time I spend on "modern" ML/DL. Note however, all the AI research I do is done on my own time, outside of the context of my employment. I intentionally chose a day job that has nothing to do with AI in order to avoid IP conflicts, so I could spend my personal time on AI stuff without concern. This means I get to set my own agenda and decide what to focus on. The downside is I'm resource limited to what I can afford to spend out of my day job paycheck. :-) Is there any future with computational Social Science, and algorithmic game theory if I want to work at those research companies? That's a good question. I kinda think that deciding ahead of time "I want to work for company X" can be somewhat self-limiting (although I have been of that mindset in the distant past. I once dreamed of working for IBM). If you really want to work for, say, Google, you probably need to focus on what you know Google are working on to a large extent. If you spend your academic time on, let's say Algorithmic Game Theory, and then after graduation find that Google have no use for that knowledge, you might be somewhat screwed from a job perspective. But if you don't limit your prospects to Google (or some small subset of companies) you might find a company that is interested in the areas you are interested in. Maybe a better plan would be to research ahead of time "are there any companies doing research in the area that interests me?" and plan on pursuing them after graduation. Another option might be to just stay in academia. Sadly, for a lot of research that isn't currently trendy in terms of commercial applications, that may well be the best bet. The other path you could pursue is doing AI research independently, supporting yourself however you choose. Maybe working a "tech job" that isn't AI related, maybe working as a bartender, who knows? Kinda depends on your background, desired lifestyle, interests, etc. Another point to consider, and I see @silicaroach has already started to allude to the same path since I started typing this, is to consider going to a Google or IBM or Microsoft or whoever, and try to gradually introduce the things you are interested in. Or heck, your area may be the "trendy thing of the day" by the time you graduate. It's hard to say. One random subjective observation to go along with the above: it seems like more AI researchers are starting to talk about "reasoning" again lately. The context is different, as it's not so much about 80's style logical inference and what-not, but there is thought/research going into how to get reasoning behavior out of AI systems. If this line of thinking continues to grow, who knows where it will take us, and what ideas will become "en vogue".