Approximately Correct

This is mostly a media hype IMO, but I have a bias as I graduated in machine learning and still work in AI. For an analysis of the state of the field and the surrounding media attention, I highly recommend this blog post by Zachary Lipton [1]. [1] http://approximatelycorrect.com/2017/03/28/the-ai-misinformation-epide…

Yes. That was the rationale OpenAI made just a few weeks ago to not release their new language models: http://approximatelycorrect.com/2019/02/17/openai-trains-language-mode…

Req'd reading: "OpenAI Trains Language Model, Mass Hysteria Ensues" by CMU Professor Zachary Lipton, http://approximatelycorrect.com/2019/02/17/openai-trains-language-mode… A salient quote from the article: > However, what makes OpenAI’s decision puzzling is that it seems to presume that OpenAI is somehow special—that their technology is somehow different than what everyone else in the entire NLP community is doing—otherwise, what is achieved by withholding it? However, from reading the paper, it appears that this work is straight down the middle of the mainstream NLP research. To be clear, it is good work and could likely be published, but it is precisely the sort of science-as-usual step forward that you would expect to see in a month or two, from any of tens of equally strong NLP labs.

Ah, I saw this last night. Here is a blog post about the article: http://approximatelycorrect.com/2018/07/10/troubling-trends-in-machine…

The tag [0] doesn't currently reference the article, it references a collection of articles. But in this instance there is currently only one article in the collection. When there are two articles with the tag, they will both be shown on the 'publishing' page, in their entirety, newest first. See [1] for an example. Redirecting would be akin to Google redirecting you to the current top result (like I'm feeling lucky, but for all searches) [0] http://approximatelycorrect.com/tag/publishing/ [1] http://approximatelycorrect.com/tag/machine-learning/

Please use permanent links, this directs to a tag, and will be invalid as soon as a newer publishing article is posted. http://approximatelycorrect.com/2017/08/01/do-i-have-to-cite-arxiv-pap…

Seems relevant: http://approximatelycorrect.com/2017/03/28/the-ai-misinformation-epide…

I thought this was about World Models paper. There maybe you can give them a little leeway: AI has more slack equating humans and machines. This paper reeks of antromorphization for the hype of it. It gets you headlines like this, but is probably detrimental to the field as a whole. Part of the Troubling Trends in Machine Learning Scholarship > In the first avenue, a new technical term is coined that has a suggestive colloquial meaning, thus sneaking in connotations without the need to argue for them. This often manifests in anthropomorphic characterizations of tasks (reading comprehension [31] and music composition [59]) and techniques (curiosity [66] and fear [48]). A number of papers name components of proposed models in a manner suggestive of human cognition, e.g. “thought vectors” [36] and the “consciousness prior” [4]. Our goal is not to rid the academic literature of all such language; when properly qualified, these connections might communicate a fruitful source of inspiration. However, when a suggestive term is assigned technical meaning, each subsequent paper has no choice but to confuse its readers, either by embracing the term or by replacing it. http://approximatelycorrect.com/2018/07/10/troubling-trends-in-machine…