Thread by @random_walker

This app by @hirevue is an example of AI whose only conceivable purpose is to perpetuate societal biases. [Thread]
theladders.com/p/26101/ai-scr…

The most questionable uses of AI seem to have the most arrogant developers behind them. businessinsider.com/hirevue-ai-pow…

Machine learning doesn’t bother with causality—questions like "how could someone’s speech patterns make them a better or worse employee?"

By rejecting the relevance of causality, machine learning has lost touch with ethics and common sense.

Dear tech world: thanks for all the correlations, but please, just stop.

Blind faith in crude tech is crippling our efforts to reverse centuries of stereotyping and discrimination.

Recently there's been a successful effort to "democratize" AI—preprints, huge datasets, pre-trained models, cloud APIs readily available.

AI in a box—no need for understanding or reflection. Sadly, these technologies are incredibly powerful yet horribly crude.

The cat is out of the bag. Anyone with no training can look at people's text/images/videos and make "predictions" & decisions about them.

Ethics education is well and good (I'm trying to do my own small part) but it might have to be combined with the heavy hand of regulation.

Research by @mrtz & others that uses causal reasoning to avoid bias in ML is super important. arxiv.org/pdf/1706.02744…
HT @angelamczhou

The @hirevue app is worse the daily AI bias stories—any signal found in body language/speech patterns *must be* the result of prejudice.

In other words, it's not that an otherwise useful app has the side effect of being biased. Rather, the app is a bias perpetuation engine.