Voices
Gary Marcus calls Astra 'extraordinarily vindicating' for neurosymbolic AI and then challenges Brockman's AGI claim
Marcus published a hot take on September 3 conceding real ground: he called it 'extraordinarily vindicating to see that a product from OpenAI explicitly creates and manipulate symbolic world models,' after a decade arguing for neurosymbolic approaches. He then attacks the rollout on two fronts, that the system is less monitorable than its predecessors, and that 'enthusiasts got an advance look while skeptics did not.' He labels his own read 'VERY tentative, pending more information about how it works.' A skeptic granting the architectural point while refusing the AGI framing is a more useful signal than either camp's press release.
↳ Follow the thread