This is a common sentiment, but seems wrong to me. It doesn't take a genius to recognise a spam or phishing email either, and yet no company has created a perfect email filter.

It's not neglect, it's a very hard problem, as is any problem where you are trying to make an inference with limited information controlled by an adversary. Amazon could spend billions on incredibly sophisticated fake-review-detection algorithms, but they still couldn't stop sellers finding real amazon customers and paying them to buy and review their products.

Neither the sentiment nor the difficulty of the problem matters. Big A gets a cut of a sale whether or not the reviews or products are genuine so they won’t do more than the bare minimum to deal with it, until their tide starts to change.