“Researchers at Meta AI (FAIR) have been attempting to train models to predict future video frames from an initial video segment for approximately 10 years.”