When isobars curve, there's a need to accelerate the air to make it curve with them. So in the N hemisphere, with the low on the left, and therefore a pressure gradient from right to left, you need more pressure gradient to make the wind follow isobars round a curve to the left than you would for the same speed of geostrophic wind with straight isobars. You need less gradient to make the air follow curves to the right.
As a result, the same gap between isobars causes a slightly stronger wind around a high and a slightly weaker wind around a low than would be predicted for straight isobars.
I have no idea why it's called a gradient wind. The terminology seems pretty dumb to me.