h is a user on social.coop. You can follow them or interact with them if you have an account anywhere in the fediverse.
h @h

Neoliberalism creating a new class divide of AI-haves and AI-have-nots?

"Why thousands of AI researchers are boycotting the new Nature journal

Academics share machine-learning research freely. Taxpayers should not have to pay twice to read our findings"

theguardian.com/science/blog/2

@h Never mind AI. The ability to even read most journals is ensconced behind The Great Paywall: titled academics and the posh on one side, the rest of us reading dumbed-down pop digest pap on the other. It’s a History Channel grade intellectual dystopia out here.

“Want to read this article? Pay only $39.99 to download!”

Mmm... I wanna eat this month instead, thanks.

@Shufei Trust me I *know*. But save for rare cases, a paper or textbook is not impossible to find if you look around on the internet.
Tackling the world's complexity is becoming a task humans can't fully comprehend, and may not fully comprehend again.

AI doesn't depend solely on papers and demonstrations of algorithms, it's also fast becoming embedded hardware. Patented hardware and productised algorithms behind APIs.

@Shufei If the trend is not stopped or at least delayed, we're going to be in a situation worse than we were when Richard Stallman identified the problem of proprietary software.

@h
I found that article lacking a bit. There's almost no mention of DATA, which is the real have / have not divide in this ML stuff.
All the big companies have open sourced the programs they use, but not the datasets that the algorithms require to function. So people train themselves on their internal tools but can't disrupt the business.
The literature and techniques in a journal amount to the same thing: useless without googbook-sized datasets.
So the Robin Hood posturing (from an Amazon employee, no less) rang a little hollow here.

@priryo Maybe it's a difficult position for an individual more than posturing, trying to push their co-workers in the right direction.

That's at least how it came across to me.

@priryo If you have something more to share about DATA, I'd gladly retoot here.