The argument against most medical advances revolves around improbable failures. Meanwhile, people die, go blind and lose their legs because they are using inferior, outdated treatments.
The device being discussed in this article is clearly preferable to purely manual methods that may need to be applied when the patient's mental faculties are impaired by a severe glucose imbalance (or even when the patient has entered a coma). Might faults in the hardware or programming result in accidents? Yes, but the risk/benefit analysis clearly favors its use.
I would add that there is a crying need for much more research into the impact of the microbiome on autoimmune diseases, such as diabetes. There is tantalizing evidence that type 1 diabetes may be curable for some individuals simply by modifying the gut bacteria.