How useful is it to play chess against the computer to improve?

What I like to do is to set up a smosition that is advantageous for me (not a material advantage but a smositional one). Then I play against the engine and try to convert my advantage. 

Wow thegreat_patzer! Full of yourself much. Pathetic!

It helps you improve your tactics!

Both have their pros and cons. The bots (if you use a pc) can predict when you are planning a tactic. This type of behavior will happen sometimes at level 3 and will become very common at level 4+ if you are only thinking 3-5 moves ahead. It will become increasingly difficult for non-tournament players which should encourage them to start thinking a few more moves ahead. The downside is that if you play the computer enough, you will start to see predictable patterns that they have which will make games very easy to win. Humans are the exact opposite. Very unpredictable, but occasionally won't be able to see if you are pulling off a tactic or not which may encourage some to play "hope chess".

As someone who's spent hundreds, if not thousands, of hours practicing against the computer, here is my two cents. You should only practice against the computer until you hit 1700 FIDE, after which you should just stop completely because past that point the cons of doing that will start outweighing the pros. What pros and cons, however, you might ask? Well, the following is my personal list:

PROS
- It improves your tactical awareness. It doesn't get any better than an engine when it comes to tactics. It'll teach you tactical awareness the hard way, punishing any oversight severely and leaving you no chance to get back in the game. You'll quickly quit hanging material when you get tired of the engine grabbing all of your stuff and blowing you off the board shortly after.
- It teaches you new tactical patterns.
- It teaches you how to defend. Being able to see all of your threats in advance, the engine will show extraordinary resilience in seemingly hopeless positions. Take any position you feel you'd convert fairly easily against a human opponent and try to convert it against the machine — you'll understand the meaning of the saying "nothing is harder than winning a won game!" Oh, and if you're at the stage where you still think every game can be won by launching a crazy attack against the opponent's king, try that against the machine and look at what happens.
- It gives you a sense of what a good move is. Though weak as it strategy-wise, the engine plays sound moves most of the time — it castles quickly, develops its pieces in the opening, avoids creating weaknesses in its camp without a good reason and punishes you when you fail to do the same. This is much better than learning from low-rated human opponents who are used to getting away with unsound play.

CONS
- You won't learn much strategy. On occasion the engine might come up with an outstanding positional idea but for the most part, it doesn't know any better than to shuffle and wait on a mistake from the opponent. In some openings even the latest version of Stockfish running on professional hardware won't have an inkling of the strategical stakes are. You'll only learn this from self-study and human players.
- It won't help your endgame technique so much. It's not that engines play the endgame poorly but they certainly play it... strangely. You can pick up a few things from playing various endings against an engine but you'll never be able to emulate its technique since, well, it doesn't have one. Everything is merely based on calculation and appears almost erratic to the human eye. A human player needs to memorize patterns that he/she can apply reliably even if it means playing second-best moves. You'll only learn that from other people.
- You'll never be as good as the machine. It doesn't matter how long you practice, it's just not happening. As explained in the pros section, you'll have a pretty decent idea of what constitutes a good move but that doesn't mean you'll be able to refute every bad move a human player throws at you, you're simply not built for that.

Hope this helps.

Thanks for the recommendation. It looks quite good, but too expensive for me, unfortunately. If I were 30 years younger and in the middle of a career instead of near retirement, I could justify the expense. But now...I have to stick to free programs like Arena. 

I can beat up the 1500 rated bot but couldn't beat a 1300 rated player (It was a good/ close game) I would say real people will help you improve more, unless you are doing the drills on chess.com those can help you. If you don't have time to play a real match, then just practice certain openings against bots.

It makes you better and it makes you humble, but losing is no fun.