Proof-of-work has a bunch of issues surrounding electricity usage, transaction scalability, and latency. It's not acceptable for a commonly-used currency to take 11 minutes to confirm transactions, nor is it acceptable for it to use more electricity than Ireland to do so. There's a bunch of research on fixing this, from various proof-of-stake systems to new consensus protocols to Nano's precomputed proof-of-work to Iota's tangle. But none of them have the track record that Bitcoin does, nor do we know if they're secure under concerted attack, nor do we know whether they'll lead to eventual centralization of power and the ability of a few actors to rewrite the ledger. I'd like to see more experimentation in this space and some actual usage with special-purpose applications before it gets rolled out by fiat across the world.

The other thing I'd really like to see is automated central banking. I don't buy the "you never need a central bank; let's just have a deflationary currency!" arguments, but at the same time central banks have a lot of problems with measurement, fairness, and incentives. I'd love to see the whole institution replaced by a computer that automatically maintains stable prices by giving money to everyone to expand the money supply and sucking it out (taxation?) to contract it, which was Milton Friedman's original idea when he came up with monetarism. To my knowledge nobody's really building this, though, and it requires solving some thorny problems in how exactly you measure price levels.