The main type of AI popular these days is tensor logic which is a digital method for finding more optimal data structures trained off big data in order to infer stuff.
Tensor based AI is only lightly based on the brain and functions well but doesn't tell you much about the human brain.
Another type of common AI is raw Turing machine AI.
Although man has investigated a lot of math over the last 10000 years man is only scratching the surface of the universe of maths. You may be able to mold a mathematical system to act like human intelligence it's still quite an unknown as to a good fit.
The next type of AI under investigation is neuromorphic AI. This is AI based on neuro electrical spiking. This can be done with analogue or digital circuits but the Analogue needs to be backed by digital for accuracy. With some tasks it is more efficient to use this method than tensor based AI.
Quantum AI finishes it off.
So far there are indicators than quantum computing might enable far more efficient data fit's.
So what about human intelligence.
Well we might be completely analogue with very minor digital aspects and that might be as far as it goes.
Better still we might be analogue mostly but rely on some quantum mechanics for organisation and optimization.
Even better still our minds may have ways to iterate quantum processes with very good error correction and nature may currently ahead of what science lightly will achieve by 30 years.