There are some problems with the definition. Humans have very high relative probability of death during their first five years, but this quickly declines and then starts to rise again. This means that humans "age" backwards for the first part of their life, which is not particularly intuitive or helpful. However, the definition is helpful in the context of understanding how our aging compares to aging in other animals.
For the first 1/3 - 1/2 of our lives our bodies are building themselves up through growth, but then you hit middle age-ish, hang up the football cleats, and start caring more about your health because "you're not a teenager anymore".
The premise that aging is defined as the probability of death feels artificial, or at least out of line with typical understandings of the word.
Old people are more likely to die simply because they are in poorer health.
So smokers or people who jump from planes may not be older compared to those who don't. We don't have a single, reliable instrument that can measure individual level aging.
Consider the phrase "He looks like he aged 20 years, but it has only been 5 years since I last saw him." Everybody understands this to mean that he looks like his health has declined in 5 years the way most people's health would decline in 20 years.
If you replace "health has declined" with "probability of death has increased", then it means roughly the same thing. It is true that the folk understanding of "health has declined" could be based on several proxies such as skin texture, baldness, etc. I think what the probability of death definition does is bring some precision to what you mean when you say "health has declined".
When my three-year-old nephew ages ten years I would hardly agree that means he's more likely to die, just the opposite. A thirteen-year-old is significantly more fit for survival.