So I also see it from the OP's perspective because...we don't need sentient AGI trillions of times smarter to solve climate change.

All we need is narrow AI that you can go look at papers demoing the results right now. And it just needs to be a little better and used to drive robots, which various startups are doing right now.

So the speculation is not "some day sentient AGI", it's "robots in the lab that work now will work a bit better, such that they can automate most tasks that robots are capable of performing now."

Why is it important for better robots to do tasks that you could use a current gen robot to do? Simple. If the robots are a little smarter with narrow AI they can do a lot greater breadth of tasks. Instead of electronics factories having robots do 80% of the work they do 100%. Instead of mines having mining equipment doing 80% of the work driven by human operators it's 100%. And so on.

This solves climate change.

It gives you several obvious tools:

​

1. It doesn't matter if cities are too close to the sea when it rises - these robots can be used to make modular subunits for new buildings, robotruck them to a site, and lift them into place. You could build an entire new city in months.
2. It doesn't matter if arable land gets scarce - self contained farms built and run by robots
3. It doesn't matter if the equator gets uninhabitable - robots go down there and get resources while people live at northern latitudes
4. We can build CO2 gathering systems and cover the sahara desert with solar panels to power them. Robots make this feasible since they do 99.9% of the labor of manufacturing, deploying, wiring up, and maintaining the panels. A sahara covered in solar is about enough energy gathering to thermodynamically reverse the last 200 years of combustion to gain energy.

The OP is right. We wouldn't be sitting helpless. Even if there is no sentience, and the tools are simply made production grade from what we already know works right now today.