I summarise work carried out on the differences in dust production by metal-poor stars, compared to canonical expectations from the Galaxy. Our conclusions are as follows. Dust may be more efficiently produced by low-mass, metal-poor stars due to their faster RGB evolution. The onset of dust production appears correlated with pulsation, which begins at slightly higher luminosities for metal-poor stars. The dust condensation sequence appears little changed for metal-poor carbon stars. Metal-poor oxygen-rich stars show marked changes in dust condensation products, with a dramatic shift occurring for globular cluster stars below [Fe/H] = -1. A lack of oxygen-rich AGB stars in globular clusters much below this metallicity currently prevents further analysis. We consider the mass-loss rate to be set by magneto-acoustic or pulsation processes on the stellar surface. Dust-driving may only become important at the highest dust-production rates where its primary role is to set the wind velocity. Future work should concentrate on obtaining gas massloss rates and wind velocities for AGB stars across a variety of dust-production rates, chemistries and metallicities to conclusively determine the role of dust in driving stellar winds.