AI Safety Debate Hampered by Imprecise Language, Experts Argue

The discussion about AI risks may be shaped more by terminology than by evidence, according to researchers who argue the field's vocabulary obscures both real harms and realistic solutions. Rather than asking whether artificial intelligence will save or destroy humanity, computer scientists and policy analysts are pushing back on the very language used to frame these questions—contending that anthropomorphic descriptions of AI systems misrepresent their nature and shift responsibility away from the companies building them.