Please commit more blatant academic fraud

The way I read this; when he says "you", it's the generic you: he's talking about the "chauvinist monsters" like himself, and it's in the context of three particular ways in which his own failings prevent him working with women.

He forms and encourages inappropriate emotional relationships with his female colleagues and it affects his ability to give criticism of their work effectively.

Now you have the right to read that a different way; but please respect that others are not just "memory-holing" misogynistic comments for some reason or another.

I'm not defending him: he definitely has a problem with his attitude to women, it sounds like it might absolutely create a hostile workplace, and it's almost certainly inappropriate to be talking about the subject in such a light-hearted manner - but this is fundamentally a mea culpa rather than a criticism of women in science.

That is true if by "AI research" you mean "applications", as opposed to contributing anything new. To contribute new knowledge what is needed is time and the resoruces to study the work of those who have contributed new knowledge before you, and to become an expert in that knowledge. In theory, that's half the job of a PhD student. The other half is creating new knowledge. Beating benchmarks is not the job of a researcher, despite what's the norm in machine learning research these days.

>> So I don't have an answer for why those coauthors went along with it.

Because the motivation of researchers in industry is money, not knowledge. Academia is capable of motivating researchers to create new knowledge. It is also capable of motivating them to commit academic fraud. But industry only understands one motive and can only offer one reward.

Basically, if your university pushes you to publish or perish, you might publish bullshit, or perish, or find yourself a niche where you can publish something that isn't bullshit. Many researchers do this, but of course you don't hear about them because they're not in the news. In fact, you probably hadn't heard about Geoff Hinton, Yoshua Bengio, Yan Le Cunn and You-Again Schmidhuber before the big boom of deep learning, because at the time they were just such small-fry researchers (compared to what they are today- I once had a look at Hinton's publication record and he was anything but small fry academically, even in the 1980's, when the lore says he was roaming the academic wilderness isolated from the mainstream; it's all bollocks). For sure, no Google was throwing millions at them at the time and the machine learning community as a whole had more or less given up on neural networks, or rather the majority of machine learning researchers where happy to leave Hinton et al and Schmidhuber to work on neural nets, while everyone else was working on Bayes nets, then decision trees, then SVMs, etc.

I digress, but what I'm trying to say is that "publish or perish" is the norm for researchers who are not motivated enough, for their own personal reasons, to make meaningful contributions. They are not the only kind of researcher and when progress happens, it comes from the other kind. But researchers in industry only have one motive, because industry offers only one incentive and is only interested in results that satisfy that incentive- regardless of how those results are brought about.

Ok, why would that not be the case? Laws? PC? Age difference? Love/biology doesn't care about social rules, and this has been shown time and again in every possible situation you could think of.

> you can sort of see how absurd it is

Huh, no I can't. What makes it different when you reverse the situation?