An investigation has been launched by US financial authorities into reports that Apple 's credit card gives different credit limits to men and women.
It is believed that the algorithms that set these limits may have been engineered to be more biased against women. Authorities have contacted Goldman Sachs, which manages the Apple Card.
Any discrimination, intentional or not, “violates New York law”, the DFS (Department of Financial Services) said.
As reported by the Bloomberg news agency, technologist David Heinemeier Hansson protested that his Apple card gave him 20 times more than the credit limit it gave his wife, even though she had a better credit score.
Later, Mr. Wozniak, co‑founder of Apple, said that the same happened with his wife, even though they did not have separate bank accounts or separate assets.
“Legal violation”
However, it seems that this is not the only case of discrimination in the world of technology. According to Mr. Hansson, creator of the programming tool Ruby on Rails, algorithms can also discriminate.
The American health group, UnitedHealth Group, is also being investigated, as there are allegations that an algorithm favored white patients over black ones.
Mr. Hansson described Apple’s credit card as a sexist program, in a tweet of his and stated that as soon as he raised the issue, his wife’s credit limit increased.
DFS declared that «it will conduct an investigation to determine whether the New York law was violated and to ensure that all consumers receive equal treatment regardless of gender».
“Any algorithm that deliberately or unintentionally leads to discriminatory treatment of women or any other protected class violates New York law”.
On Saturday, the investment bank said to Bloomberg: “Our credit decisions are based on the creditworthiness of the client and not on factors such as gender, race, age, sexual orientation or any other discrimination prohibited by law.”.
No one is yet sure what is really going on, but there is a suspicion that an unintentional bias has entered the system.
This could be because when the algorithms, they were trained on a dataset in which women actually posed a greater financial risk than men. This could cause the software to offer lower credit limits for women in general, even if the underlying assumption doesn't hold true for the population at large.
Alternatively, the problem may lie in the data currently fed to the algorithms. For example, in married couples, men may be more likely to take out large loans using their own name alone, rather than jointly with their wives, and the data may not have been adjusted to take this into account.
