I don’t think a behavioral test is enough. There needs to be additional large scale integrations.
I’ll try to come up with an example for this…
How do we know something is alive? It’s through the combination of several fields of observation: on the surface level we observe certain behaviors, biologically, we learned that living things have genetic material.
Thanks to this, we know that mold is alive, while it may have been considered radically different in the past, since its behavior is so different than animals’. We can then expand our knowledge to include bacteria as a living thing, and exclude things like volcanos or stalagmites as being alive, while a primitive culture may have been inclined to say that the volcano is “angry” and the stalagmites are slowly growing creatures that feed on water and rock.
In some cases you need to know something from the inside out to be able to determine its nature with certainty.
I think in the case of AI and consciousness, we would need to understand the phenomenon of consciousness. And that’s not gonna happen for another decade at least.
And I think that to properly study something like this, a scientist would need to be part philosopher, and not just the type that shows up at a lab, ready to run experiments and collect statistics. They’d need to integrate conclusions from fields like anesthesia, attention, sleep, neurosurgery and neuropathology, evolution of biological neural networks and so on.