No. She was accused of doing shoddy research and was fired for that reason, not her point of view, therefore evaluating the research is the only relevant issue.

No. She was accused of doing shoddy research and was fired for that reason, not her point of view, therefore evaluating the research is the only relevant issue. I welcome this second opinion on the paper and in fact this article changed my opinion. I really am blown away by the energy cost of building these models, and would love to hear that the numbers in the paper are just way way off. I realize that very few models like this are built, but as Gebru points out, that also introduces bias and the problem of verifying the integrity of the process. I don't have to agree that NLP is racist, or that capitalism is racist to find value in this paper. And it certainly is true that what many take to be inevitable in technology is in fact a choice, so attacking and questioning fundamental assumptions is a very important part of moving into the future.