|
|
 Originally Posted by oskar
Here's something mildly interesting:
Any phrenologic examination based on images requires topographical analyses. Topographical analyses based on images without knowing the exact angle of light source is a shitshow even in multi-scopic analyses. If They did have a proven methodology to analyse topography based on monoscopic sources without knowing the angle(s) of the light source(s), I'd want to know everything about it because I know some people who would lose their shit if this was possible. Forget about gaydar, that's a billion dollar algorithm if it exists.
I don't know about all this, but I imagine the AI is categorising certain facial features and measurements like the size of the nose and the distance between the eyes and whatnot. So I wouldn't be surprised if the AI did possess some gaydar - like abilities.
I'm also not surprised at seeing a tenuous link being drawn between this and the other tenuous link between a weak effect of political orientation and attractiveness in political candidates.
More generally, before criticising someone for inferring causation from correlation, it seems important to understand whether the correlation actually exists in fact or whether it exists mainly in the tattered logic of that person's own head.
|
Reply With Quote