well when you say it like that lol...
The human part of it, it's just brain tissue, sure there would be issues if it were to be described as "sentient" but I don't really think that's the case necessarily for just stuff being grown in a lab. These aren't fully formed brains, just pieces.
And on the AI front, I'm pretty skeptical that this is the "world shattering" importance that its backers ascribe to it, but it's undeniably now an important part of the global computational infrastructure, and will be for some time.
SO in the big picture, since (LLM-type) AI's are basically just neural networks, and if there's some way that brain tissue can be "trained" to produce something that mimics neural network outputs... Then I'd say this is worth exploring in a limited (scientific) sense, keeping in mind ethical constraints of course.
And yes there's a "mad scientist" quality to it that I can't say I don't find exciting. 😃