Values distant from most other values are?

Outliers. An outlier is an observation unusually far from the bulk of the data, such as a recorded adult height of 250 cm. It may be a measurement or data-entry error, or a genuine but rare event.

A standard rule is Tukey's fences: flag values below Q1 − 1.5 × IQR or above Q3 + 1.5 × IQR, where IQR = Q3 − Q1. If Q1 = 20 and Q3 = 30, the fences are 5 and 45, so 52 is an outlier; box plots draw such points individually. The rule |z| > 3 is also common, but outliers inflate the mean and standard deviation that z uses, so robust versions use the median and median absolute deviation.

Interviewers ask what you would do with one. Do not delete it automatically: check whether it is an error; if real, consider robust methods such as the median, Huber loss or tree models. In regression, separate outliers in y from high-leverage points in x and influential points, measured by Cook's distance.