When it comes to extra terrestrial life, past Earth experience cannot be reliably used as a good predictor for the future. We need to map possible future scenarios and apply logic and reasoning.
This is like ants arguing that life outside their ant hill will likely have 6 or 8 legs because all the life they've seen so far have had 6 or 8 legs.
Maybe we cannot be certain that AI can exist, but it's certainly reasonable to predict that AI will surpass us, given the fundamental principles involved. Much like people who predicted that the Earth was round before it was circumvented. We've only seen a fraction of what AI can do, but there is no fundamental barrier to AI cognition.
Also, the only conclusion you can draw from SETI is that no one nearby is actively sending a signal too Earth. We aren't able to pickup "background" EM activity beyond a ~20 light years - which is only a handful of stars. Moreover, the notion of ET omnidirectional broadcasting so inefficient, it would be difficult to explain (thus unlikely).
My guess is that we are surrounded by quiet AI observation nodes, either in elongated orbits, or extra-solar transients. Probably near impossible to detect. There doesn't seem to be any great harnessing or storage of stellar solar outputs in the galaxy, so either their efforts do not involve large amounts of energy - or they've found some other means to capture it (maybe that's what dark matter is).