We describe an empirical model of climate based on multiple linear regression of the century long global surface temperature record that considers variations in radiative forcing due to GHGs, tropospheric aerosols, the solar cycle, ENSO, volcanic aerosols, the strength of the Atlantic Meridional Overturning Circulation (AMOC), as well as climate feedback and ocean heat export. We use this model to suggest cooling caused by major volcanic eruptions has been overestimated by about a factor of 2, due to prior neglect of variations in the strength of the AMOC. The representation of volcanic eruptions by global climate models will be critiqued and implications of our analysis for geo-engineering of climate will be described. Next, we illustrate how the cantilevering of aerosol radiative forcing and climate feedback results in divergent estimates of future temperature, for regressions constrained to match the same century long temperature record. These results will be related to output from IPCC climate models, which exhibit remarkably similar behavior as our regression analysis. Finally, we show better quantification of aerosol radiative forcing and climate feedback for the contemporary atmosphere is needed to provide reliable forecast of future global warming.