I smoke; most of my family smokes, around half of my acquaintances. As best as I can tell:

External messages (warnings, PSAs) have a negligible effect on people who already smoke. The psycho/physiological considerations are much more a factor than how graphic a warning message is.

Instead, they should be doing this study on people who don't smoke. The real goal is prevention.

Personally, I'm a programmer so I don't worry about warnings anyway, only errors.