So as of today, we don't really have any kind of simple metric like "IQ" that we can apply to AI systems, and even if we did have the metric, at this point we wouldn't really know it's range, whether to expect it to be a linear function or not, etc.
That said... does it seem clear that we are making some kind of progress? I'd say yes, but with the admission that that's somewhat subjective. In particular, even though we know that current AI systems are getting better at doing whatever they do, we don't really know if that meaningfully counts as progress towards AGI. It is possible, for example, that LLM approaches will hit a ceiling in their development, and that the much feared "GPT 17" really won't even be a thing. And it may likewise turn out that the progress made on LLM's doesn't turn out to contribute much at all to whatever the eventual mechanism behind AGI turns out to be.
Is that the way I expect things to play out? Not sure. I have a hunch that LLM's will prove to be insufficient to qualify as AGI on their own, but that they may well serve as an element of a modular / hybrid system that includes LLMs, reasoners using deduction, induction, and abduction, evolutionary algorithms, and FSM-knows what else.