Can AI stop thinking that accepting help is a sign of weakness?