I don't know for certain, but the two major faults of Diamond are first, that his framework for anthropology is perpetuating beliefs that are antithetical to modern anthropology, and second, that he is extremely resistant to following up with new advances in anthropology. On the latter point, for example, Diamond is still a believer in Clovis-First hypothesis, despite that having been discredited for so long that I learned about it in grade school decades ago.
On the first point in particular: there's a general thesis of anthropology which is properly termed unilineal model (and in more colloquial terms, Civilization-style tech tree), which views civilizations as existing on a spectrum from less to more advanced. In essence, GG&S reduces the question of "why did Europe conquer the New World and not the other way around" as "because Europe was more advanced" (and then go on to try to explore why Europe was more advanced), but never trying that hard to justify why "was more advanced" was the answer to the first question. By contrast, the consensus view of modern anthropology is that technological development is heavily adapted to the context of its culture, so that (for example) the Incans don't develop carts because carts are useless in their mountainous environment.
Once you go beyond the idea that "more advanced" is a qualitatively useful metric of comparison and actually start demanding to see the causal links in how specific technologies are necessary to achieve the claimed outcomes, it starts to become clear just how badly Diamond whiffs it. Literacy is a fun example: Diamond has to work around the fact that the conquistadors were largely illiterate, and they were facing bureaucratic, administrative imperial states that ran on record keeping, requiring at least proto-writing [1]; really, how different were the two empires in terms of literacy? Another fun example I like to bring up is comparing Tenochtitlan (the Aztec capital) to Toledo (the Spanish one). Tenochtitlan had over 100,000 inhabitants, universal primary education (for both boys and girls, even!), running water, and public amenities like zoos and history museums; the Spaniards had... none of that.
[1] And there's a case to be made that Incan quipu were an honest-to-god writing system.
On the first point in particular: there's a general thesis of anthropology which is properly termed unilineal model (and in more colloquial terms, Civilization-style tech tree), which views civilizations as existing on a spectrum from less to more advanced. In essence, GG&S reduces the question of "why did Europe conquer the New World and not the other way around" as "because Europe was more advanced" (and then go on to try to explore why Europe was more advanced), but never trying that hard to justify why "was more advanced" was the answer to the first question. By contrast, the consensus view of modern anthropology is that technological development is heavily adapted to the context of its culture, so that (for example) the Incans don't develop carts because carts are useless in their mountainous environment.
Once you go beyond the idea that "more advanced" is a qualitatively useful metric of comparison and actually start demanding to see the causal links in how specific technologies are necessary to achieve the claimed outcomes, it starts to become clear just how badly Diamond whiffs it. Literacy is a fun example: Diamond has to work around the fact that the conquistadors were largely illiterate, and they were facing bureaucratic, administrative imperial states that ran on record keeping, requiring at least proto-writing [1]; really, how different were the two empires in terms of literacy? Another fun example I like to bring up is comparing Tenochtitlan (the Aztec capital) to Toledo (the Spanish one). Tenochtitlan had over 100,000 inhabitants, universal primary education (for both boys and girls, even!), running water, and public amenities like zoos and history museums; the Spaniards had... none of that.
[1] And there's a case to be made that Incan quipu were an honest-to-god writing system.