There's an important difference between a tool having a formal structure that we learn to use skilfully and a tool just being a kind of learned skill. Is our understanding of a mathematical operation an understanding of how to apply a rule or just a behavioural regularity that we look at abstractly?

Or (just maybe) what you think you know so far contains a misunderstanding and your ‘hard problem’ is a warning of an unseen contradiction.

I would sooner assume that I’m vainly trying to explain something I’ve misdescribed than posit a parallel universe of reified sensations or sentient electrons.

The behavioural data is well worth thinking about carefully but I don't think the headline result is technically demonstrated

For some, this is a contradiction: to say something anthropomorphic about a model is the same thing as making a claim about its 'real' intelligence. But rather than accept the equivalence and deny the language (which is futile), we should point out the conceptual error

For some, it informs their ideas about moral responsibility and therefore their political opinions and therefore their voting behaviour and so on. But what's in scope of what you mean by "how you live"?