It doesn't necessarily mean exactly the same biases since you could deliberately implement the opposite bias in some of those cases. But it suggests that such algorithms will consistently have overall a similar set of biases.