Bamber & Aspinall 2012 made an expert elicitation of ice sheet experts. They shows this table in their supplementary information:
This table shows that on average then most experts expect that ice sheet loss more than doubles when you double warming. Semi empirical models (SEMs) generally assume linearity. Arguably this suggests that SEMs may be overly conservative. However, ice sheet modelling supports the linearity assumption in SEMs (e.g. Bindschadler et al. 2012).
Another very interesting plot from Fettweis showing a strong relationship between Greenland surface mass balance from an RCM and global warming in the encapsulating GCM.