Smart contracts have a major hurdle called the 'Oracle problem'. In summary, it is hard to get data from the real world into the blockchain. Even if you put it there, why should anyone trust you. A way around this is to get many credible parties to input the 'state' from their perspective and then blend the inputs. Once you have a data stream that you can trust (relatively) on the blockchain, you can make non-custodial trustless (or trust-minimized) smart contracts (bets, predictions, derivatives, etc.) based on the original variable in the real world.
All things that work just fine without the blockchain.
Blockchains only solve one problem: when two or more parties don't trust each other, and do not have the protections of the legal system (trying to be anonymous or conducting illegal acts). They will forever remain a solution in search of a problem.
Smart contracts do solve one additional problem: for contracts where performance is measured on purely objective, measurable criteria, execution is swift - there is no scope of disputes hence legal costs are reduced. You also don't need a 3rd party to oversee the execution of contracts, reducing management/transaction costs.
where performance is objectively measurable on the blockchain, if it's in the real world, you introduce trust issues again. How could one trust the measurement was objective in the real world?
Yes, definitely in the current format, smart contracts don't obviate trust issues. It is proposed that a 3rd party oracle will be used to verify the criteria is met for execution. Oracles can be incentivized to build credibility.
For example, in the case of insurance contracts, the oracle can be sensors on the car which can make decisions about who's at-fault and accurately assess damage.
There are plenty of countries where protections of the legal system isn't guaranteed, even if you're not trying to be anonymous or conduct illegal acts.
Bloomberg made boatloads of money by being the go-to provider of financial data infrastructure. Blockchain is littered with startups making that bet, but now quite serious players are making that bet too.
Regardless how far up the hype curve you think blockchain is, it's interesting that AP is leveraging its reputation to commit to a project in this emerging tech.
Idea is that oracles are that bridge between the physical and the blockchain worlds, and it's these interfaces that are always the problem for the blockchain idealists. Playing that bridge role relies on being a trusted source so people believe you won't just run away with their funds... a startup doesn't start with the reputation needed to pull that off, but someone like AP does. So it's an interesting competitive advantage for someone like AP to do this.
So we can create derivitives to gamble with, of course. Come on, it's not the 1990s anymore. Instead of betting your uncle $100 that Trump is going to win it's now possible to use synthetic dollars to earn arbitrage profit by rebalancing the automated market makers who are supplying liquidity to all of our uncles bets on who will become president.