From the point of view of classical economics, AI increases the productiveness of capital, and certain specialized kinds of labor (e.g. creating and improving AI).

While this increases the total output of society, it also increases inequality.

We already have in place mechanisms to deal with inequality, namely income tax and welfare. As the total output of society grows, we will be able to afford more generous kinds of welfare, until eventually we probably settle on a negative income tax scheme.

Note that a correctly designed welfare system wouldn't result in lots of people living on welfare, but rather everyone working fewer hours.

So I imagine that the future could be quite good, as long as our methods for redistributing wealth are able to keep pace with technological change (and there are some great examples to follow in the Western world, e.g. Scandinavia, Australia and NZ)