I do not think a linear trend extrapolated that far into the future is reliable. The slope appears to have changed several times over the decades. A Gompertz fit, as posted by others, appears to be a better fit, but it looks to flatten the curve too much.
I was taught that the gompertz or S curve is used to represent the way an event happens. A classic example is expenditure on a construction project. Costs are low at the beginning, (design, approvals etc,) accelerate in the middle as main construction takes off and slows down with fiddly finishing work at the end. We used it in doing budgets for capital expenditure programmes.
The curve is also usual when looking at the annual melt of the Arctic Ocean as a whole. The melt speeds up as temperatures rise and slows down when temperatures cool down. The second reason for the gompertz curve being appropriate as a representation is that when looking at an individual sea, no matter how early it melts out, when sea ice drops below a certain amount, the percentage of the remaining sea ice lost each day remains very much the same, and therefore the absolute decrease in daily sea ice declines, creating the classic gompertz end of graph shape. But there is no reason to suppose that to be the case with immediate future years ice loss in the Arctic.
At the moment, the increase in CO2 atmospheric concentrations is accelerating. This is likely to accelerate the increase in atmospheric temperatures.
If, as expected, pollution decreases due to public opinion and decreased use of coal, this is also likely to accelerate the increase in atmospheric temperatures,. albeit temporarily.
Summer Sea ice extent minimum is still at 50 % of the 1979 value.
Winter Sea ice extent maximum is still at nearly 90 % of the 1979 value.
We are nowhere near the tail end of the event being progress to an BOE. So why should future years annual losses in Arctic sea ice decline in line with the tail end of a Gompertz curve?
The only thing that makes any sense to me is to start with the idea of what is going to happen to AGW in general and AGW in the Arctic in particular in the immediate future, i.e. the next 5 to 10 years. And that looks pretty grim to me. How you stick that into a curve is beyond me. I don't have a few cray computers to help me out, either.