How AI Could Turn Medicine into a Commodity

In March 2020 musician/programmer/copyright attorney Damien Riehl and a programmer named Noah Rubin changed music copyright forever. The two of them worked together to create an algorithm that generated every musical melody possible (within certain bounds of western music, but for all intents and purposes, every melody possible). The pair then saved those melodies to a hard drive, claimed the copyright, and released it. This effectively made musical melodies un-copyrightable.

What this algorithm essentially did was turn melodies into data sets. Riehl argued that under copyright law “numbers are facts” and facts have little to no copyright.

What would happen if we applied this approach to medicine development — an industry reliant on exclusivity through patents?

Turning Medicine into Data

Imagine an AI that could create every possible combination of chemical compounds.

Just like melodies, there are finite drug combinations working within the bounds of molecular interactions. So if an AI could go through every possible combination, you could make the same argument that Riehl and Rubin did: If AI discovered every possible drug combination, is a drug patentable?

Why This Matters

When a drug company patents a medicine, this typically gives them the exclusive right to manufacture and sell that medicine for 20 years (shocker: medical companies find ways to extend these patents). This exclusivity means Pfizer can charge the consumer whatever they feel like for things like, say, an EpiPen. The idea of a patent makes sense. It encourages companies to innovate and it makes a lot of sense in a lot of cases. Medicine isn’t one of those.

Commodities

What’s a commodity?

A commodity is a basic good: They’re inputs in the production of goods and services rather than finished goods. There’s also little to no differentiation in that good regardless of who produces it.

Take water. Water is water. You can’t own the idea of water. It’s water. You can bottle water. You can source it cleaner, or more efficiently. Fiji can market better, handle supply chain more efficiently, and have higher profit margins than Dasani, but Fiji can’t stop Dasani from selling water.

What is water? A molecular structure, good old H₂O, there’s no other combination of molecules that makes water. Just H₂O.

Medicine as a Commodity

Let’s bring this thinking to medicine and an AI that comes up with every possible drug combination. Along with turning all of medicine into a dataset, it also, theoretically, does something else.

Back to water real quick. Human’s need water. There isn’t a base alternative. Water is the perfect solution to the problem of hydration.

In theory, if an AI creates every possible drug combination, there will be a “perfect” solution to each medical problem.

For example, somewhere in the massive dataset that would be created, there should exist a molecular combination that results in a drug that perfectly cures, say, the flu, with no side effects.

Once this AI discovers that perfect combination, the idea of medicine completely changes. We now have the correct solution to specific medical problems like the flu, cancer, or diabetes. It’s no longer about trial and error or incremental improvements. There is a “perfect” drug.

So Then What Happens?

Then we’re faced with a new question. Can a drug company claim a patent on these perfect solutions if they’re found in a database of molecular combinations that an AI created? Or, does that perfect drug become like water?

And I know what you’re thinking, “but poor Pfizer, they haven’t made enough money yet, how could they ever make a profit if they lost their monopoly.”

Welcome to free market capitalism Pfizer.

Drug companies would have to be introduced to the concept of competition. One company would no longer have the monopoly on a drug for the next 20 years. Instead, they’d have to come up with ways to manufacture, distribute, and sell that drug better than other companies. Oh yeah, and price competitively.

This Changes Costs

Manufacturing and distributing a drug isn’t the part that results in exorbitant prices. Aside from the “because we can because we have a monopoly” part of high drug prices, R&D is what really drives up costs.

The process of creating a new medicine involves discovery (our new AI friend does this part), clinical trials, and getting regulatory approval from the FDA; and while a drug that is successful becomes profitable to the nth degree, drugs that fail can result in millions of dollars lost (in simplistic terms, fancy accounting will typically offset that).

So AI solves discovery, but there’s a valid argument, and probably a factual one, that this commodification of medicine takes away the incentive for companies to run trials.

Here’s where I’ll lose some of you. The government could step in and take over trials.

The Federal Government

I know, I know, boo Uncle Sam getting involved in the markets. Why use taxpayer dollars to create medicines when a private company would have done it instead?

Here’s the thing. You, the taxpayer, fund A LOT of the R&D that goes into creating the medicines you can’t afford.

The National Institutes for Health (NIH) plays a major role in funding biomedical research. In fact, between 2010 and 2019, the NIH, and by extension, you, funded 99.4% of the drugs approved by the FDA (that ran you $187 billion by the way).

A study from Bentley’s Center for Integration of Science and Industry published in JAMA Health Forum showed that the amount that the NIH invested per drug approval was comparable to that of the reported investment by the biopharmaceutical industry.

You’re already subsidizing drug development that companies are then patenting, claiming they own, and are charging you whatever they want for it.

If this AI solution drastically cut the costs of discovery, and even assisted in trial design and predictive modeling, and the NIH and non-profits started running trials, this could not only save you money, but also do something that is very rarely said about the government, create efficiency. The pipeline of drug discovery, trial, then FDA approval all of a sudden becomes a streamlined process done entirely in-house by the government with no external involvement. There could be several other beneficial side-effects of this.

The government doesn’t have an incentive to get you addicted to a pill,

for example (looking at you Sacklers).

This is Very Good for the Consumer

So what exactly does a pharmaceutical company become now? A water bottle company.

There would still be PLENTY of money to be made in the manufacturing, distributing, and selling of drugs; but now these companies would have to actually innovate and actually compete, and competition means lower prices.

This isn’t a novel concept, even in the pharmaceutical world. Cost Plus Drugs, for example, is already doing this. They even tell you exactly how they do it.

The example they provide on their website is this:

“For a 30 count supply of 400mg of Imatinib”, a game changing cancer drug, they “take the manufacturing cost of $31.20, add a 15% markup of $4.80, and $5.00 for pharmacy labor. A total cost of $41.00.” (before shipping and taxes), and that’s without insurance.

The retail price of a 30 count supply of 400mg of Imatinib is $9,616.30

Why? Because the manufacturer of Imatinib can, so they do. So, why isn’t Cost Plus Drugs doing this for Ozempic or any other revolutionary new medicine that’s come out in the last few years? Those silly little patents. Cost Plus Drugs is only able to manufacture drugs that have a “generic” version, or a drug that is no longer protected by a patent.

What’s the difference between a generic of a drug and the name brand? At the molecular level of the active ingredient, the commodity, nothing.

In Conclusion…

Artificial intelligence is going to do a lot for the world. Some good, some bad, some revolutionary. An artificial intelligence that could commodify medications would have an unmeasurable benefit on society and Big Pharma would have to come down to earth and play the capitalism game with the rest of us.