4 ms·
I have wasted my time watching these long winded tortuous and even hour long videos that set out to explain to us dumb data scientists that algorithmic bias is
by random314 5y ago
I have wasted my time watching these long winded tortuous and even hour long videos that set out to explain to us dumb data scientists that algorithmic bias is a problem. And there isn't any algorithm or proof or equation in these garbage laden soup of words. And if there is a good example of bias, it is always in the dataset in these long winded diatribes.
However, I must say they are quite consistent in their introductory paragraphs. They start by acting dumbfounded that people don't understand that algorithmic bias has already been demonstrated and proven to be a problem, so we should be talking about how to solve this problem instead of "pretending to be not aware ". Except they won't directly link to the work, simply mention the authors name - typically Gebru. And we must Google, read the paper, find that its garbage and waste our time.
FWIW, I am a person of color.
- criddell 5y agoHave you read Cathy O'Neil's book Weapons of Math Destruction? If so, what did you think of it?
- AlanYx 5y agoFunny you mention that book. It's cited in a huge proportion of AI fairness papers, especially papers by academic authors who do not have a technical background, and I can't figure out for the life of me why. The book itself is more polemic than academic. One of the heuristics I actually use when reviewing papers in this area is to check whether that particular book is cited without any references to page numbers, like a mantra rather than an academic citation. This seems to correlate fairly highly IMHO to low quality work with few, if any, interesting original contributions.
- darawk 5y agoTotally agree. I found the total lack of equations, data, and experiments in the paper this thread is about extremely telling in this regard as well. Lots of vague assertions, zero rigor. I think all the points being made in this paper evaporate rather quickly when subjected to any sort of serious formalism.
- criddell 5y agoHave you read the book? If so, what did you think of it (even as a polemic)?
- random314 5y agoInstead of asking someone to read an entire book as a response to a paragraph, which is just an extreme version of what happens to be the talking point in every discussion about "the bias is in the algorithm, not just the data", why dont you quote relevant paragraphs from the book here? Responding with - have you read this or that is not the fait accompli you are imagining. It doesn't make you look more informed than the OP. You actually do have to provide an argument instead of have you read this or that.
- criddell 5y agoI wasn't asking anybody to read a book. I haven't read it myself. Have you? If so, do you think it's worth reading?
- noptd 5y agoPrecisely, and it's strikingly similar to a religious pattern of discourse. "You know the dogma so don't resist. Embrace it and live your life accordingly."