Operating Systems or Artificial Intelligence

[jk123]

I want to choose the area of ​​specialization - operating systems or artificial intelligence.

On the one hand, I'm interested in operating systems and low-level programming. But this area seems to me to stop. It seems to me that nothing new happens here, "all have been invented". This article (though written long ago) confirmed my thoughts:

http://doc.cat-v.org/bell_labs/utah2000/

On the other hand, artificial intelligence seems to be something very interesting and even exciting. There are many interesting areas - machine learning, machine vision, pattern recognition. Artificial intelligence is associated with robotics. I heard "Around 2025, being a robot programmer will be in demand as much as being an app programmer is today. Study lots of A.I. Start early". But I have friends who say that artificial intelligence - the area is quite boring, mainly related to the meticulous study of algorithms or "primitive selection of parameters for algorithms and preparation of input data for them."

[AtomicMaster]

Operating Systems are always under development, and there are tons of unsolved and improvable problems.

Both areas are difficult and interesting, with tons of unsolved problems; regardless of which one you pick, you won't have a dull day...

[Popcorn Sutton]

My area of specialization is pattern recognition particularly. Yes, it is a very exciting field and believe it or not, there are jobs out there for this type of thing (I started mine last week). What I've come to find is that you really need to apply yourself and let your potential employers know exactly how valuable it is and the potential for income is substantial if it's done properly. If you get a job with artificial intelligence, there's really only 3 known ways to work with it. One is through rules, the other is through statistics, and the last is through both. I think that people like using both for the most part. I tend to use statistics and logic mostly. It's fun.

 

It's mostly an R&D type of job though. You really can't expect to get in the field and develop a smart program overnight, so don't say that you can. From what I've seen, people with PhD's have spent their entire career developing toys basically, but they demonstrate a point at least. That's why I chose pattern recognition really. The application is useful for a lot of purposes.

[AtomicMaster]

If you want to look at areas of unsolved problems in OSes, just for example, look at hypervisors and virtual networking, which, there are projects right now that don't even do IPv4 yet with something like 1.2 million lines of code... Trick there is providing a virtual network that works with real networks, and works like a real network, only it is not, and runs as a process...

 

Just an example of an unsolved problem in OSes