Maybe, but my point is that if it's this bad at understanding chess, which a fairly simple and objective domain, why should we have confidence it has understanding in other domains where it's harder to prove wrong answers wrong? We aren't going to be able to come up with specialized plugins that fix its mistakes in every conceivable domain. The amount of trust people are putting in the current models seems to indicate that they think it's roughly at a human level of understanding, which doesn't seem to be the case. And if we are actually trying to solve the problem of general intelligence, we can't just sweep things under the rug and say "plugins will fix it".