Can you improve playing a computer?

I recently bought an chess app on my ipad, and since everyone always says to play against stronger opponents, would playing against that regularly help improve my skill? It's not some 3000 rated computer or anything, probably about 1800-2000 and you can change the difficulty from 1 - 10, what do you think? 

Why not?

I think sure can.. but not a blitz game with computer

The problem with playing against a computer is twofold.  First is that while computers are tactical giants their positional "understanding" is limited (although getting better).  As a result it simply doesn't play like a human would.  The second problem is that there is no way to evenly "dumb down" a computer to become a good opponent at your level.  The program instead makes periodic blunders and then goes back to playing at it's usual strength.  This is again not the way humans play.   All that said, if (and it's a very big IF) your app has an opening "book" associated with it then playing can be useful to learn openings.  Otherwise it's not all that productive.  Sometimes it's fun though, especially if humans have been beating you regularly, to set the program at a low level where it gives you a material advantage and then just beat the snot out of it.  So fun, yes; productive, maybe not so much.

I practice against the computer quite often.

I think I'm at a point where I play much better against it than I play against humans and that's mostly psychological.

I know it never makes a mistake so I must not allow any inaccuracy in my calculation and sometimes think for 20 minutes per move. (Computer is forced to play 1|10, the app allows me to continues when my time is out)

When playing humans, even if I have that much time, I often feel tempted to not let them wait much longer for my moves than they took for their own, since I somehow feel that's impolite.

So my "best game ever" is one where I managed to stay at material-equality for 41 moves against Stockfish. I usually surrender when I can't keep it up anymore.

Btw. I have to reject the claim that they don't have "positional understanding". At least for Stockfish I'd argue that it's hard to be better there. If this was the case, it wouldn't be able to outplay about anyone with same material in the endgame. I often thought that it's pawn-structure doesn't look that good compared to my own, but in the end the apparent weaknesses weren't real as he could make sure there's no way to exploit them by insane calculation-depth.