It's an interesting thought. I have had a similar thought about AI and consciousness while tripping. But there are many pros and cons that just end up in speculation. Which is interesting but not very productive?
what basis would you recommend? Given the rate that chatbots engaging in making stuff up you can't just ask them. At least as humans we can take the substance ourselves to verify there's something happening. I think there will always be an insurmountable barrier to deciding if computers can be conscious, at least in our lifetimes. We're not even sure other humans experience similar consciousness.
I suppose we could simulate tripping in that fashion but it would seem easier to affect a distributed activation function or other parameter to the simulation.