3 ms·

Algorithms can have bias, too. Like for instance, in the last election, Google removed the more negative expressions that would be suggested alongside politicia

by mtgx 9y ago

Algorithms can have bias, too. Like for instance, in the last election, Google removed the more negative expressions that would be suggested alongside politicians' names. That wasn't just the Algorithm™ -- it was Google deciding that. There are likely hundreds of such "manual" decisions within its search algorithms.

zaarn 9y ago
Algorithms are generally incapable of bias. What an algorithm does is a perfect reproduction of it's instructions plus it's parameters. The actual bias is either by the designer by making an algorithm that does not fit the intended use properly or by using flawed parameters. Also known as; garbage in, garbage out. The pipe in between doesn't care if it's garbage. In your example, this was a decision by a Google engineer or c-level, not the algorithm. The algorithm did as it was told to in exactly the way it was designed without bias or error.
- ForHackernews 9y ago
  > Algorithms are generally incapable of bias. This is nonsense unless you define-down "bias" in such a way as to mean "performing contrary to the deterministic result of its implementation" (which seems to be what you're doing, and I agree is impossible). A more sensible definition of bias might be something like "preferentially favouring one outcome over another for reasons unrelated (or even contrary) to the core purpose of the software". Imagine a vending machine that when prompted generically by a user saying "I'm thirsty" always provided sugary soda instead of bottled water. Modern machine learning algorithms are extremely complex and opaque, and depending on the dataset they're trained on, and the features they're trained to focus on, they can very easily wind up just encoding human biases. https://www.theregreview.org/2017/10/10/cramer-bias-computer-algorithms/ https://www.theregreview.org/2017/10/10/cramer-bias-computer...

by mtgx 9y ago

zaarn 9y ago

Algorithms are generally incapable of bias. What an algorithm does is a perfect reproduction of it's instructions plus it's parameters. The actual bias is either by the designer by making an algorithm that does not fit the intended use properly or by using flawed parameters. Also known as; garbage in, garbage out. The pipe in between doesn't care if it's garbage. In your example, this was a decision by a Google engineer or c-level, not the algorithm. The algorithm did as it was told to in exactly the way it was designed without bias or error.

ForHackernews 9y ago
> Algorithms are generally incapable of bias. This is nonsense unless you define-down "bias" in such a way as to mean "performing contrary to the deterministic result of its implementation" (which seems to be what you're doing, and I agree is impossible). A more sensible definition of bias might be something like "preferentially favouring one outcome over another for reasons unrelated (or even contrary) to the core purpose of the software". Imagine a vending machine that when prompted generically by a user saying "I'm thirsty" always provided sugary soda instead of bottled water. Modern machine learning algorithms are extremely complex and opaque, and depending on the dataset they're trained on, and the features they're trained to focus on, they can very easily wind up just encoding human biases. https://www.theregreview.org/2017/10/10/cramer-bias-computer-algorithms/ https://www.theregreview.org/2017/10/10/cramer-bias-computer...