I’m curious if y’all are paid like real interns, $15 an hour, or if you are doing some godawful Amazon Mechanical Turk thing where you get paid one and a half cents for each website you label as “hate.”


UHRS is a crowdsourcing platform that supports data labeling for various AI application scenarios. Our vendor partners connect us with people—who we refer to as ‘judges’—to provide data labeling for us at scale. All UHRS judges are under NDA so your data is always secure.

The point is it is not machine learning, that is similar to the “click all images of a bus” – that is “data labeling for AI.” Obviously, the bus, train, traffic lights, crosswalks – that is “neural network training” presumably for Google’s self-driving cars. It’s actually really fascinating how it works – it is not at all like you think, it is completely unintuitive.

To get all philosophical about it, it is not that there is no “strong artificial intelligence” it is that there is no strong natural intelligence. Not only is there no logic at the base of human cognition, it really is just a soft, wet, analog computer and instead of NAND gates it’s all almost fractal like spider webs of neurons.

How do kids learn to read? Do they understand the underlying structure of the grammar? No, they just repeat what they hear. That is the real secret of human language – no grammar is logical. (Long diatribe about the artificial nature of late Latin grammar deleted.)

In any case, our friends at UHRS are actual humans, not bots, and they were sent here to decide whether or not I’m a Good Liberal Democrat or a bad Trump Republican, thus, “extremist hate speech” that gets censored by the two search engines that haven’t already censored my blog.