Overall I think the point has merit - much of what has been done in AI today is at least on the right path to artificial general intelligence (AGI). "The most important parts of AGI have already been achieved by the current generation of advanced AI large language models." But this article in particular raises only one good objection and three straw man objections. The good objection is 'scepticism about metrics for AGI'. Better tests are needed (but based on our experience with humans, may not be forthcoming). In fairness to the authors, most objections to today's generative AI do fall under one of the three straw man headings - devotion to an alternative theory of AI, belief in human exceptionalism, or concern about the economic consequences. But none of these comes even close to showing that we have not achieved general AI (nor is that usually the authors' intent).
Today: 0 Total: 14 [Direct link] [Share]
