I would like to know the controls on this study.
Did they control for overall caloric intake? For percent of body fat?
If they did not, this might simply be a marker for more calories, more risk.
Check out Steven Milloy's Junk Science Judo if you're interested in a nice exposition.
I thought diet side had zero calories.
They didn’t control for fat or calories consumed; the artificial sweetened consumers had the highest BMI of the the three subgroups (no drinks, sugar sweetened drinks, artificially sweetened drinks). The authors suggest that these could represent former sugar sweetened drink consumers who then switched to artificially sweetened to reduce caloric intake. A better study would use a larger population and eliminate other risk factors, in other words, do skinny people who drink tons of diet or sugary soda (there are lots of them) have an increased risk of Type II diabetes.