Can artificial intelligence be biased?