“. . . how do we make sense of a machine that passes the bar exam but flunks preschool? “
– Will Douglas Heaven
MIT Technology Review studies opposing views on the state of intelligence in large language models. At the centre of the debate is whether they primarily work by reasoning or recall.