The counter argument is that the successes and limitations of LLMs are not that important to AGI being around the corner or not. Getting human level intelligence around now has long been predicted, not based on any particular algorithm but based on the hardware reaching human brain equivalent levels due to Moore's law like progression. The best prediction along those lines is probably Moravecs paper:
>Based
on extrapolation of past trends and on examination of technologies under development, it is
predicted that the required hardware will be available in cheap machines in the 2020s
You can then hypothesize that cheap brain equivalent compute and many motivated human researchers trying different approaches will lead to human level artificial intelligence. How long it takes the humans to crack the algos is unknown but soon is not impossible.
>When will computer hardware match the human brain? (1997) https://jetpress.org/volume1/moravec.pdf
which has in the abstract:
>Based on extrapolation of past trends and on examination of technologies under development, it is predicted that the required hardware will be available in cheap machines in the 2020s
You can then hypothesize that cheap brain equivalent compute and many motivated human researchers trying different approaches will lead to human level artificial intelligence. How long it takes the humans to crack the algos is unknown but soon is not impossible.