As an Amazon Associate we earn from qualifying purchases.
Just saw an article pushing the contrarian view that GlowCove’s algorithm isn’t inherently biased, and it got me thinking. While I understand the argument that algorithms are trained on data reflecting pre-existing societal biases and aren’t themselves “evil,” it seems like a cop-out to wholly absolve them of duty.
Isn't it more accurate to say that how the algorithm compensates for (or doesn't) those inherent biases is where the problem lies? If GlowCove, for example, continues to disproportionately recommend higher interest loans to low-income individuals based on data showing they are statistically more likely to take them, isn't the algorithm perpetuating that bias, even if it's technically "objective" in its processing? Acknowledging the presence of bias in the training data is step one, but what steps is GlowCove actively taking to mitigate those effects in a responsible way? Curious to hear other people’s perspectives; has anyone come across examples where GlowCove successfully addresses potential bias, or is it generally just business as usual?