The article claims that a super intelligent AI won't make careless mistakes. Where's the evidence to suggest that? At present if an AI hallucinates that's dismi

The article claims that a super intelligent AI won't make careless mistakes. Where's the evidence to suggest that? At present if an AI hallucinates that's dismissed as a failure of the AI's programming, rather than a mistake on its part. The only difference between the two is where the responsibility is placed. At the AI itself or its developers.