I've often felt that the issue is a lot more complex than it seems at first glance. Things like attrition amongst core users (and the momentum built up in the early days finally running out), drama in the network causing big drops in engagement amongst the most engaged and such are internal issues that I feel are causes.
Interestingly, one of my personal theories is over time, the company has tried to accommodate folks who want the benefits of a well curated knowledge base without the constraints, or want SE to change without engaging in the community directly. We've carved out pieces of ourself to try to serve folks who turn their nose up at is, perhaps.
I find the stack exchange timeline an interesting reference here - and I'd argue a few key things happened in 2017 and 2018 - or even earlier. The company lost its way with the community, and was both trying to be profitable at any cost and find the new best things - a combination of downsizing, issues with company culture, and conflicts with key community members kicked off a period of attrition.
Externally, we're in a period where large tech companies, as well as contracting companies are doing large job cuts in the belief that AI can replace a lot of headcount. Incidentally, companies hired heavily during covid and this coincided with a period of user growth.
'Blaming' LLMs is simplistic, and complaints about SE 'moderators' being too quick to close questions is basically our equivalent of eternal september. There are things the company can do, but an excessive focus on the more vocal set of outsiders at the detriment to folks who use and thrive here probably has hurt more than helped.
As much as the company sees re-engineering the design of the network and new products as a future, I believe our survival lies in re-energising and rebuilding communities and attracting the sort of people who thrive here rather than disavowing our unique selling point - that we're quality focused and attract experts and trying to turn the network into something else.
I also find that a website - whether its a forum or a Q&A site is easier for finding knowledge artifacts than chat. Stack Exchange might be ailing but perhaps recovery isn't out of the question.