We studied this question here: arxiv.org/pdf/2403.00499
I think that much of the discussion around the intelligence of models, current or future, in fact revolves around consciousness or the ability to have subjective experience, rather than objective performance measurements on any external task.
arxiv.org
