This is based on # of conversations started. That's it. It's just bad data to compare AI use across professions that spend vastly different amounts of time sitting at a computer not in any meetings.