I'm betting the economics of recovering gravitational potential energy don't really work out in most use cases.
I realize there may be some over compensation in play, but unless the counter weight's weight is actual adjusted for each load, there should be something to work with.
Any weight difference due to meatsacks in the elevator will be relatively small compared to what a drum full of concrete weights. Going off the math in TFA, and assuming your average meatsack weighs 75kg, lifting one of them 15 meters up (so, roughly, to the top floor of a 6 story building) stores about half a smartphone battery worth of potential energy. So you probably want to be lifting them much more than 15 meters - and be packing your elevator cars full of them - if you want to pay off the cost of the fancy regenerative kit in any reasonable amount of time.
There's also a big difference between doing it in an elevator, which operates intermittently, can't be scheduled to match grid supply and demand, and works with relatively light loads, and doing it with a dedicated system that can match the grid and use much heavier loads.