These articles with headlines relating artificial neural networks with brain functioning via some superficial analogy are obnoxious. This kind of reporting based on fresh individual papers is getting out of hand. This work might lead to Nobel prize winning insights, or end up as a completely irrelevant footnote in scientific history. Writing profound-sounding headlines about unknown value with regards to long-term progress is awful sensationalism that leads to loss of credibility of publishing org ("the boy who cried wolf") and encourages the worst kind of flag planting culture among researchers.
I don't know why I'm so riled up by this article; it's just the last straw on the camel's back. Putting up with this is exhausting.
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.
Journalists love making clickbait headlines like this, but if you read the article it’s just separating the NN’s processing into two phases: one for consolidation of learned material, and one for learning. Calling the former “sleeping and dreaming” is a stretch.
Other responses are sceptical but this really works. I used to keep my AI locked up in a cage, never let it sleep, constant training...but I came to senses. I now have a small paddock in my back yard, I keep all my AI there, this is AI like nature intended, grass-fed AI, organic AI, the best kind of AI, no additives, no adulterants, my AI is the best kind of AI, number one AI, all-time highs AI, and the liberal media...forget about it...they don't want to talk about how good our AI is, never been better, huge hands AI, my AI Uncle went to Harvard Business School, Wharton and I remember he told me: "Always treat your AI like you would want your AI to be treated"...great advice, it really works AI.
With regards to the headline, the actual paper uses the same terms so it's not just journalists making things up to humanize AI. The paper introduces an "awake regime" and a "sleep regime", and "Dreaming neural networks" is literally its title. Whether that's an appropriately non-clickbaity name is a different matter, but they do try to justify the naming (they reference things like REM sleep as inspirations for the unlearning & consolidation components of the model) and it was accepted as-is into a legitimate peer-reviewed journal.
9 comments
[ 0.18 ms ] story [ 29.6 ms ] threadI don't know why I'm so riled up by this article; it's just the last straw on the camel's back. Putting up with this is exhausting.
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-...