Traditionally, the circulating water gets colder as it reaches the arctic circle and submerges, since cold water has a higher density than warmer water.
But: melting ice in the arctic releases freshwater. This reduces the salinity at the surface of the Northern Atlantic. Water with lower salinity is lighter. Therefore, this water no longer submerges, and the circulation is slowed down.
Indications of the process having started (edited) before humans could be responsible for it (few hundred years ago).
Approximate salt content of the Atlantic Ocean: 11,000,000 billion metric tons
It was whether we can produce enough to bring the fresh water that is being added from melting ice up to normal Atlantic salinity.
If we’re going to hasten our demise, we might as well try sulfate aerosols or something with a more global effect. We’d still end up killing ourselves, but at least it wouldn’t be quite so much like trying to get a sailboat to move by blowing on the sails.
There must be some quantity of salt we can dump a year to raise salinity. We should be rigorous, and find out what that quantity is. If we can change a climate, we can change an ocean.
So, to raise salinity by .01%, we'd need 1268 years of normal production (plus some when you exclude sea-water extraction).
[0] http://www.wolframalpha.com/input/?i=(volume+of+the+world%27...
[1] http://www.wolframalpha.com/input/?i=31956+cubic+miles+of+sa...
[2] http://www.saltinstitute.org/salt-101/production-industry/
Yes, we probably can re-salinate a flow to attempt to alter its dynamics. We'd also have altered its salinity profile in ways which have repercussions on e.g. cod fisheries, or krill, or other things. Thats the problem with complex biological systems with lots of interrelated parts: they don't reverse with simple tweaks because once you change one part, you change lots of parts in unexpected ways.
(I did once study this stuff back in the 80s and was very bad at it, but what I recall of even two-actor predator-prey relationships is that they are remarkably complex, cyclical, and do not just return to the same stable place if you alter them trivially. I believe this is probably even more true for physical systems like weather and climate)