5 ms·
Indeed, one could consider machine learning an authoritarian technology if given power over individuals. AI's fundamental problem is it winds-up being a "result
by joe_the_user 8y ago
Indeed, one could consider machine learning an authoritarian technology if given power over individuals. AI's fundamental problem is it winds-up being a "results oriented" approach where individuals characteristics are weighed by black-box systems and the individual is judged without any recourse or even any exact idea what the criteria is.
- trevyn 8y agoWell, we’re judged by human brains every day, which are even more opaque.
- sgt101 8y agoThis is a sorta standard rejoiner.. but other humans should have to account/take responsibility for their decisions, and part of that process is that other humans should judge individuals as individuals which almost by definition machine learned classifiers cannot. ML is great for dealing with massproduced artefacts, or standardised processes, but it's just not the right tool for reasoning over domains that are not full of approximately similar things. That the Chinese government (and Google, and Facebook) think that humans are in that category speaks volumes about them.
- pixl97 8y agoIn the long term human brains are input and computationally limited.
- joe_the_user 8y agowe’re judged by human brains every day, which are even more opaque. In modern society, the decisive things determining life-direction are determined by human ostensibly using objective criteria. Educators have conditions for grades, bosses write progress reports, courts use etc. This situation is term bureaucratic society and the rule of law. Authoritarianism is one kind of interaction that we all know - when an individual can issue orders without recourse. There's everything from abusive bosses to dictators here. It's a common part of human relations too, just a less desirable side.