It seems like criticality is sufficient, bot not necessary, for TAI, and so only counting criticality scenarios causes underestimation.
It seems like criticality is sufficient, bot not necessary, for TAI, and so only counting criticality scenarios causes underestimation.