Not sure. AI doesn’t seem to be as good yet at (synthetic) data scarce areas like radiology (!) or even in maths at say coming up with road maps to difficult proofs. But yes maths and CS communities may be more honest too.
AI’s rapid gains in math/theoretical CS get lots of attention, but these fields aren’t uniquely automatable—or even the most exposed. Some of the visible “panic” may simply reflect that math/theory researchers are being largely open about using AI and candid about how to respond.