The new series of artificial intelligence models are able to undertake more complex tasks, though risks remain in regards to privacy, data safety and human error.