I too believe ceramics will precipitate from liquid metals rather than dissolve. That's why I put "I don't see how to introduce atom pairs of MgO" and suggested to dissolve the ceramic-forming metal, then introduce oxygen, and quench before the ceramic coalesces.
Maybe a melt is the wrong direction. Accumulative roll bonding might disperse a complex in a metal at a reasonable temperature. It wouldn't be bonding here, but the same process.
Or you might insert the precursors to the solute in different sheets of the base alloy, then bond the sheets finely by accumulative roll, and once the precursors are very close to an other, heat moderately to let them diffuse and react with an other.
If the base metal is very unreactive, as gold is, maybe CO would dissolve. At least, it won't precipitate. Advantage of all gases.