Non-insulin-dependent Diabetes or Type-2 diabetes is a harmful condition but yet is very common in the population. According to WHO, The number of people with diabetes has risen from 108 million in 1980 to 422 million in 2014.
However, diabetes is not a recent cause of concern. James V. Neel, a geneticist at University of Michigan Medical School proposed the "thrifty gene" hypothesis in an attempt to explain "why certain populations and subpopulations in the modern day are highly prone to diabetes mellitus 2".
Neel suspected that Type 2 Diabetes likely had a strong genetic basis (way before genome sequencing was known) and considering it has such a harmful effect, why has it been favored by the process of natural selection. Neel suggested the resolution to this problem is that genes which predispose to diabetes (called 'thrifty genes') were historically advantageous, but they became detrimental in the modern world. In his words they were "rendered detrimental by 'progress'".
Example from Wikipedia,
According to the hypothesis, the 'thrifty' genotype would have been advantageous for hunter-gatherer populations, especially child-bearing women, because it would allow them to fatten more quickly during times of abundance. Fatter individuals carrying the thrifty genes would thus better survive times of food scarcity. However, in modern societies with a constant abundance of food, this genotype efficiently prepares individuals for a famine that never comes. The result of this mismatch between the environment in which the brain evolved and the environment of today is widespread chronic obesity and related health problems like diabetes.
Further research by Neel failed to provide reasonable evidence to support the original hypothesis. As such, different modifications to the hypothesis were proposed, of course, along with their renewed criticisms.
In modern day research, large scale GWAS studies suggest a polygenic model of causation for Type 2 Diabetes, where multiple genes contribute to the disease pathology (with smaller effect sizes) than a single gene. Further, it appears that these contributions come primarily from changes in regulatory landscape of the cells (>90% of the GWAS hits for T2D are non-coding).