Only if it changes laws of physics. Which I suppose could be in the realm of possibility, since none of us could outthink a ASI. I imagine three outcomes (assuming getting to ASI) - it determines that no, silly humans, the math says you’re too far gone. Or, yes, it can develop X and Y beyond our comprehension to change the state of reality and make things better in some or all ways. And lastly, it says it found the problem and solution, and the problem is the Earth is contaminated with humans that consume and pollute too much. And it is deploying the solution now.
I forgot the fourth, that I’ve seen in a few places (satirically, but could be true). The ASI analyses what we’ve done, tries to figure out what could be done to help, and then suicides itself out of frustration, anger, sadness, etc.
Over time, it is. It’s eliminating the source. In Terminator, Matrix, and others they say that the AI took a split second to act, but our AI doesn’t have those connections. It’s working with what it’s got.