When AI meets archives : testing commercial general-purpose LLMS for technical and content-related description