There is real disagreement over this issue among AI insiders, including among much less prominent people. It makes it hard for outsiders to take the position that AGI is definitely possible or definitely impossible, especially in the medium term, based on the evidence available.