Security Notes: How AI Escrow Is Creating Global Trust

What comes first, the payment or the service, the money or the merchandise? It is best when they happen simultaneously, but this is not always possible. The seller says, “Pay me now and trust me to deliver later.” The buyer says, “Deliver now, and trust me to pay later.”

If payor and payee are acquainted and have a shared history, they also may have trust, and then the transaction is less of a problem. In many trades, this dilemma is resolved by paying half before, and half after. Other solutions are based on a pay-as-you-go plan, but when payor and payee are mutual strangers, suspicions emerge.

If the deal is substantial and mutually advantageous, the parties go for an expensive escrow solution, where typically a law firm holds the payment until it is satisfied that the delivery of the purchased service or merchandise took place. But the overhead and expense of a full-fledged escrow account makes it impractical for small transactions. Alas, small transactions are the majority.

Our world is so village-like in cyberspace that myriad opportunities to connect commercially are awaiting us, ready to pump prosperity on a global scale. There are some organizational solutions. Fiverr, for example, connects buyers and suppliers from all around the globe and plays a conflict-resolution role. Typically, credit card companies offer some simple conflict-resolution interventions, and indeed some card users take the quality of this conflict-resolution service into account.

But none of this is sufficient when we enter the digital-money world, where payor and payee may transact without the umbrella of a custodian network that charges an arm and a leg. Here, the new solution to the question of trust and transactional conflict resolution is based on the computing power that changes life everywhere else: artificial intelligence, or AI.

Conversational AI packages are already well developed. They are being adjusted to run effective conversations between buyer and seller, and to clarify issues in conflict. Parties with a public footprint are analyzed for trustworthiness and credibility. The detailed dialogues with the parties are AI-processed to prepare a case summary that includes all relevant information. If the summary leads to a clear answer, the AI conflict-resolution package declares that.

But if the case is questionable, the plan is for the summary to be submitted to paid “judges” randomly selected from a given list. The identities of the conflicting parties are hidden from the judges, who give an action recommendation accompanied by a self-measure of confidence that indicates the degree to which they are at peace with their decision. This “wisdom of crowds” is then reduced to a case decision. Over time, the AI package will learn from the crowd and will not need them.

The parties agree a priori to abide by the decision of the AI package. The buyer then passes the funds in digital coins to the AI operator, which in turn disposes of the funds according to the conflict-resolution verdict, subtracting a pre-agreed service fee.

Such automated escrow services allow for two complete strangers to do business with a global reach. They would not be confined to some network’s rules, nor surrender metadata to a powerful collector. This emerging capability would lift payment into the role of a powerful peace agency.

History teaches that commerce between societies is a precursor for peace. Strangers that connect over the LeVeL payment protocol will connect on a human level soon enough. BitMint money is designed for this purpose. BitMint and other developers are rushing to develop this particular AI capability, untouched by the concern that AI could subjugate humanity.

—Gideon Samid gideon@bitmint.com