When Berkshire Hathaway executive Charlie Munger said, “Show me the incentives, and I’ll show you the outcome,” during a speech in 1995, he wasn’t talking about health care, but he might as well have been. Companies in any industry are influenced by how the norms and structures around them contribute to their success, and in health care, the incentives often aren’t set for the good of the end user—the patient.

Consider these incentives from the perspective of the health plan–provider relationship. While there are plenty of people in health care doing genuine work to serve patients, like in any industry, there are also bad actors. Because of the subjectivity that is baked into health care’s billing and coding policies, those bad actors are incentivized, and given latitude, to game the system. That ambiguity allows a doctor with less-than-pure intentions to, for instance, upcode a procedure they shouldn’t without pushback from the institution (or individual) paying out the claim.

The vendor stack

At the same time, the complexity that underlies these policies has created an environment where vendors can sell software and services that help health plans sort through claims and make sure they’re being paid out correctly. Theoretically, that’s a perfectly valid value proposition, except that the vendors that health plans hire to find errors aren’t incentivized to work in an efficient way.

Vendors typically are paid based on how many errors they find, so rather than addressing root causes that lead to bad billing practices, they’re motivated to maintain the status quo. That means they are often unwilling to do things like share the errors they find, since that would lead to fewer errors in the future, and thus lower payouts.

It’s the same inertia that has led to the claims administration process to rely on old systems for processing data because there aren’t strong enough levers motivating the organizations responsible for finding errors to maximize their chance of arriving at the right decision the first time. The loser in all of this is the health plan, which ends up paying for a vendor stack of six or seven companies using old technology to solve a problem poorly.

The need for change

These market inefficiencies are well-established, yet the norms have remained for a couple of reasons. For one, it’s nearly impossible for humans to legislate out all subjectivity and complexity from the system that allows for bad actors to benefit. Plus, it’s difficult to change incentives that have been in place for so long when they’re benefiting established players.

A new solution is needed—something or someone who can serve as a neutral third-party arbiter, one that’s not influenced by any incentive other than arriving at the truth efficiently. Humans, and the organizations they work for, will always be unable to achieve true neutrality. So, we need a better partner: AI.

AI as a neutral arbiter of truth

AI can be trained on rules. When those rules are clear, well-defined, and avoid introducing bias, it ensures that AI does not deviate from its mission to find the truth. This means that AI can ingest and interpret the complex policies that govern health care claims adjudication accurately, consistently, and quickly.

By creating standardization and removing subjectivity, bad actors aren’t able to skirt a rule for their own benefit, and there is less opportunity for perverse incentives to skew outcomes. Providers, for instance, will all code the same procedures in the same way, because there will be no ambiguity that allows for deviation. In this alternative reality, there’s also less of a need for a bevy of vendors to sort through claims and find errors, since the uniformity of process will lead to fewer errors in the first place.

It has been well demonstrated that the current system, built on manual processes, doesn’t lead to good outcomes for people or institutions who pay for care, and thus, is bad for the health care system as a whole. What’s needed is a neutral third party to help plans, providers, and vendors all simplify claim adjudication. AI can be that neutral arbiter, and everyone who touches the health care system will benefit.

Tim Wetherill is a physician executive.