There is a lot of talk about artificial intelligence reading scans. Here is where I land on it.
Everything in medicine that uses AI still needs a human being responsible for the result. There is no such thing as AI making the medical decision. It does not exist, and it should not.
I use these tools the way I use a stethoscope. They extend what I can perceive. Automated detection can flag a lesion I then confirm or dismiss. Computation can measure what my eye cannot. Language models can help me phrase what a structure does.
But not one of those outputs leaves as an answer until a trained human has checked it, signed it, and taken ownership of it.
The value of AI is that it makes a careful reader faster and more thorough. The danger is pretending it can be the reader.
A machine can assist the judgment. It cannot carry the responsibility.