The reality is the AI engineers that are working on AI autopilot systems for self-driving cars still tell their users to keep their eyes on the road. AI medical or doctor apps that suggest possible conditions still tell their patients to consult a real doctor. AI trading systems that auto-trade and analyse the markets are also limited and don't account for multiple risk points and still need supervision by the multiple traders and the lawyers using AI to sift through historical cases to save time still need supervision under a human lawyer.

Where we're going is assisted AI; obviously not a full on replacement as the scare stories created by the AI hype squad in this thread after the first reactions of this tool.