The Prisoners' Dilemma is a problem from Game Theory and Mathematics which, in its essence is about trust.
Two people have just robbed a bank together - they're apprehended by the police a few days later and each is separately taken in for questioning where they are offered the chance to confess.
They're both offered the same deal - testify that your friend was the one who robbed the bank and you're free to go OR stay silent and we're going to charge you with something, even if we can't charge you with the bank robbery.
If each prisoners betrays the other then neither one is set free and both get punished; and if both keep their lips sealed then they both only get a minor punishment. But if only one prisoner betrays the other then the prisoner who stayed silent takes all the blame and the other is declared innocent.
There's a debate on what the optimum solution is for each participant but it depends on the greater context of the prisoners and their relationship (will they rob banks together again, how bad is the betrayal, how good is the honesty) but as a thought experiment the parallels with business should be obvious, Even with strong contracts and good relationships there's often an incentive for one party to betray another and the incentive has to be managed - the prisoners' dilemma has never been confined entirely to the university campus and has been applied liberally.
Smart contracts offer a technical solution to any digital version of the prisoners' dilemma - they don't just optimise the problem - they solve it.
If the interactions are digital then Prisoner A and Prisoner B can write a contract before they rob this bank - the rules of the contract (written in computer code) are that if either party betrays the other, then the other party will automatically also betray the first as well. This removes any incentive for anyone to betray in the hope of doing so alone; because the contract is held across the network and not by either A nor B then both can be sure that the contract exists in the form that it was agreed on and the lack of trust is not just pushed up stream. No additional party is required to execute the contract because it executes as conditions are met (i.e. one party betrays) and the agreement cannot be repudiated in any form, and so can be trusted indefinitely.
One way of looking at this conclusion is that wherever there is a Prisoners' Dilemma in business there might be an opportunity for smart contracts on a decentralised ledger.