Papers of the day   All papers

Language (Technology) is Power: A Critical Survey of “Bias” in NLP

Comments

Hanna Wallach: Alright, people! New paper alert!!! (w/ @sulin_blodgett, @haldaume3, and @s010n) Prepare for a thread! https://arxiv.org/abs/2005.14050 First, the incredible @sulin_blodgett surveyed 146 papers (yes, really!!!) analyzing "bias" in NLP systems. (1/8)

5 replies, 353 likes


Annette Zimmermann: On why #NLP research (and #AI research more generally) needs #philosophy + #linguistics: https://bit.ly/30doWuk I really benefitted from reading @aylin_cim's seminal work on this+a fascinating new paper by @sulin_blodgett @haldaume3 @s010n @hannawallach (https://arxiv.org/pdf/2005.14050.pdf) https://t.co/vLOUHka2fs

1 replies, 20 likes


Ted Underwood: We badly need a better frame for these conversations than the word “bias” (which inadvertently reaffirms the neutrality it’s trying to question). This paper goes a better direction by admitting that we’re making normative choices about “allocational” and “representational harms.”

1 replies, 12 likes


Thomas Davidson: Interesting paper and thread. In short, research on bias in AI doesn't sufficiently engage with social science or discuss the meaning or social implications of bias. How much of the problem is due to the constraints of the medium - conference papers with strict page limits? 1/2

1 replies, 9 likes


Lauren Klein: “We further find that these papers' proposed quantitative techniques for measuring or mitigating "bias" are poorly matched to their motivations and do not engage with the relevant literature outside of NLP.”

0 replies, 7 likes


Maria Luciana Axente: A great paper and thread on bias in NLP @GolbinIlana @ShabnamRashtchi

0 replies, 6 likes


Mark Riedl wears pants during video calls: 👇 this is a doozy

0 replies, 6 likes


Kasumi Yamazaki: Blodgett et al. (2020) Language (technology) is power: A critical survey of “bias” in NLP https://arxiv.org/abs/2005.14050 https://arxiv.org/pdf/2005.14050.pdf 🔹A survey of 146 papers analyzing “bias” in NLP systems 🔹Proposing 3 recommendations that should guide work analyzing bias in NLP systems

0 replies, 5 likes


Rishi Bommasani: @srush_nlp I found this to be a lovely read, hats off to @sulin_blodgett for surveying the 146 papers on social bias or fairness at the time. To appear at ACL 2020, @hannawallach has a nice thread describing the paper. https://arxiv.org/pdf/2005.14050.pdf

0 replies, 4 likes


HotComputerScience: Most popular computer science paper of the day: "Language (Technology) is Power: A Critical Survey of “Bias” in NLP" https://hotcomputerscience.com/paper/language-technology-is-power-a-critical-survey-of-bias-in-nlp https://twitter.com/hannawallach/status/1266378580218589184

0 replies, 3 likes


ML and Data Projects To Know: 📙 Language (Technology) is Power: A Critical Survey of "Bias" in NLP Authors: @sulin_blodgett, @s010n, @haldaume3, @hannawallach Featured in PTK #41 https://bit.ly/ptk41 Link: https://arxiv.org/abs/2005.14050

0 replies, 3 likes


Iason Gabriel: One key idea: our aim should be to engineer conceptual categories that mitigate conditions of injustice rather than entrenching them further. We need to deliberate about which social practices and structures—including linguistic ones—are valuable before we automate them (from AZ)

0 replies, 3 likes


arXiv in review: #acl2020nlp Language (Technology) is Power: A Critical Survey of "Bias" in NLP. (arXiv:2005.14050v1 [cs\.CL]) http://arxiv.org/abs/2005.14050

0 replies, 2 likes


Annette Zimmermann: “If you can do things with words, you can do things with algorithms”: http://dailynous.com/2020/07/30/philosophers-gpt-3/#zimmermann My #publicphilosophy piece on expressive & representational harms in #NLP #GPT3 #philosophyofAI

0 replies, 2 likes


arXiv CS-CL: Language (Technology) is Power: A Critical Survey of "Bias" in NLP http://arxiv.org/abs/2005.14050

0 replies, 1 likes


Hanna Wallach: Alright, people! New paper alert!!! (w/ @sulin_blodgett, @haldaume3, and @s010n) Prepare for a thread! https://arxiv.org/abs/2005.14050 First, the incredible @sulin_blodgett surveyed 146 papers (yes, really!!!) analyzing "bias" in NLP systems. (1/8)

0 replies, 1 likes


Content

Found on May 29 2020 at https://arxiv.org/pdf/2005.14050.pdf

PDF content of a computer science paper: Language (Technology) is Power: A Critical Survey of “Bias” in NLP