Weather is a chaotic system, but often chaotic systems have longer-term trends that are possible to observe. For example, if you pour sand into a pile, then you can fairly accurately predict the shape of the cone that it will create, but the exact pattern of bounces for each grain is impossible to predict as it depends on the exact position, shape, and location of every grain that it hits on the way down and a tiny error in any of these will magnify to a huge error after a few bounces. It's a chaotic system that has macro-scale effects that can be predicted.
Weather is a chaotic system with very similar properties. The longer the timescale, the easier it is to predict. Predicting the average temperature difference between summer and winter, for example, is much easier than predicting the temperature tomorrow.
To give a simpler example, if I toss a coin 100 times, I'd expect you to be able to tell me, with a fairly small margin of error, how many times I will roll heads. I wouldn't expect you to be able to guess what the result of any individual toss will be more than about half the time.
If you think that predicting weather and predicting climate are similar problems, then I'd encourage you to read up a bit on chaos theory.