I’ve honestly wondered if autocorrect machine learning has picked up on poor autocorrect choices, causing a reinforcement loop, causing it to continue getting worse