I believe there is already a few 10s of nm thick layer of indium in production processes for that purpose (or to stabilize bonding?).
Copper metallization works (usually with a nickel layer and a silver electroplate on top which uses substantially less Ag than a nuclear reactor of the same net power), but is presently considered cost prohibitive process-wise as it adds a few steps. The current process being rolled out is silver plated copper paste (each copper bead has a 10-30% silver coating) as well as some techs for laying it down more consistently (aligned screen printing where the line never crosses two layers of mesh, and once it matures, depositing it with laser).
Silver consumption may briefly increase if the above facilitate a switch to heterojunction or topcon cells, but as soon as one of the 3 tandem chemistries being tried per day shows itself as a winner the voltage will increase and the fingers will get significantly smaller (to the point where Al may not cause a shading problem). At that point the shills will start whining about a 100nm thick layer of lead containing molecules while ignoring that it's less lead than your average tuna.
