Why Heideggerian AI failed and how fixing it would require making it more Heideggerian