The Quiet Code HIV, Artificial Intelligence, and What We Are Still Not Talking About

A few days ago, I was doing something almost all of us do now without thinking much about it. I was scrolling.

Photographs. Politics. Advertisements. Friends. News. Videos designed to make us stop for five seconds before moving on to the next thing.

Then I stopped.

The image showed HIV floating inside what looked almost like a glass sphere. Across the bottom were the words: “New discovery shows HIV can be silenced permanently using its own hidden genetic code.”

It was the word permanently that caught my attention, although probably not for the reason whoever created the post intended.

Science rarely gives us words like permanently that easily.

So I started reading.

Underneath the social-media headline was something much more interesting than the headline itself. Researchers have been studying an RNA molecule produced by HIV called the antisense transcript, or AST. In research published in Science Advances, scientists increased expression of AST in CD4+ T cells taken from people living with HIV who were receiving antiretroviral therapy. In laboratory experiments, that additional AST made latent HIV substantially harder to reactivate. The researchers describe AST as a biological molecule capable of reinforcing HIV latency and possessing potential relevance to future curative strategies. They did not demonstrate that HIV had been permanently silenced in a person. PubMed

That distinction matters.

But I kept thinking about something else.

I kept thinking about the information.

The code.

The blood samples. The laboratories. The researchers. The people living with HIV who participated in studies. The people who agreed to give another vial of blood because maybe somebody would learn something from it. The doctors who kept records. The scientists who sequenced viruses. The universities that stored information. The public institutions that funded research. The communities that lived through an epidemic long before anybody could imagine that a tiny piece of RNA hidden inside the virus might someday become part of a conversation about keeping HIV asleep.

Science does not usually begin when we announce the discovery.

It begins much earlier.

And that brought me directly back to another article I had just written.

Read “If the Machines Can Remember, Why Can’t the System?”

In that article, I was looking at the negotiations over the World Health Organization's proposed Pathogen Access and Benefit-Sharing system, or PABS. The language is technical, but the question underneath it is not.

When countries share pathogens and genetic sequence information that allow scientists somewhere else to develop vaccines, medicines or diagnostics, what happens to the people and places where that knowledge began?

Who remembers them?

Who shares the benefit?

Those negotiations are still unfinished. WHO Member States concluded another negotiating session on September 18 without completing the PABS annex. On September 25, countries meeting at the United Nations again called for the work to be completed. The annex is intended to connect the rapid sharing of pathogens and sequence information with fairer access to the vaccines, therapeutics and diagnostics developed from them, and WHO says the current process is headed toward consideration at the World Health Assembly in May 2027. World Health Organization

The HIV research I was reading about does not suddenly fall under PABS simply because genetic information is involved. They are different scientific and legal conversations.

But I could not stop seeing the connection.

Because underneath both stories is the same question.

What happens after information becomes valuable?

We Talk About Data as Though It Has No Childhood

I have been thinking about that sentence for several days.

We talk about data as though it has no childhood.

A genetic sequence appears on a computer screen and eventually becomes another piece of information moving through an enormous scientific system.

But it came from somewhere.

Before it was data, there was biology.

Before the biology became a sequence, somebody collected it.

Sometimes there was a person sitting in a clinic.

Sometimes there was a frightened family.

Sometimes there was an outbreak.

Sometimes there was a community that had already lost people.

Sometimes there was a researcher working in a laboratory that did not have nearly the resources of the institution that would eventually turn that information into something commercially valuable.

And sometimes there was simply someone living with HIV who agreed to participate in research because there was still so much we did not know.

By the time information reaches the other end of that journey, everything looks different.

There may be patents.

There may be investors.

There may be pharmaceutical companies.

There may be licensing agreements.

There may be stock valuations.

There may eventually be a medicine.

We become very good at remembering everybody standing near the finish line.

We are not always as good at remembering everybody who helped us leave the starting line.

And now artificial intelligence is entering that story.

Maybe We Are Asking the Wrong Question About AI

Right now, AI seems to arrive in almost every conversation wrapped in anxiety.

What jobs will disappear?

Who will misuse it?

Will students stop learning?

Will artificial intelligence become too powerful?

Will people become too dependent on machines?

What happens if governments lose control of it?

Those questions should be asked. Guardrails matter. Accountability matters. Privacy matters. Human beings should remain responsible for decisions that affect other human beings.

But I wonder whether fear has become so loud that it is beginning to drown out another conversation.

What could happen if we deliberately used this technology to make life better?

Not someday.

Now.

AI does not need to become human to be useful to humanity.

It can help us see relationships across quantities of information that no individual human being could realistically hold in one mind. It can compare enormous bodies of research. It can look for patterns. It can connect one finding with another finding published years earlier. It can help scientists ask questions they might not have known to ask.

And it can help us remember.

That part interests me enormously.

Imagine a scientific system in which the origin and history of biological information travels alongside the information itself.

This sample came from here.

This sequence was contributed here.

These researchers identified this characteristic.

This community participated in this study.

This institution funded this work.

Another laboratory added something ten years later.

A researcher somewhere else found a relationship nobody had seen before.

An AI system helped scientists recognize another connection.

That connection eventually contributed to a therapy.

The therapy produced enormous value.

Suddenly the history of the discovery does not disappear simply because the science became complicated.

I am not talking about attaching a royalty meter to every scientific paper or making researchers ask permission every time they think.

Science cannot operate that way.

I am talking about memory.

Provenance.

A record of contribution sophisticated enough to reflect the way twenty-first-century science actually happens.

Because the problem is going to become much larger as AI becomes more capable.

Imagine an artificial intelligence system analyzing millions of viral sequences, decades of medical literature, protein structures, treatment histories and laboratory observations.

It discovers a relationship nobody had noticed.

Scientists test it.

It works.

A biotechnology company develops it.

Clinical trials follow.

Eventually there is a treatment.

Now tell me where that discovery began.

With the company?

With the AI?

With the scientists who validated the finding?

With the researchers who generated the data?

With the laboratories that uploaded the sequences?

With the patients whose samples made those sequences possible?

With the countries where the biological material originated?

Maybe the truthful answer is all of them.

Our systems are simply much better at determining who owns the product at the end than remembering everybody who made the knowledge possible at the beginning.

That is something we are not talking about enough.

AI Can Accelerate Extraction, Too

There is another side of this conversation that we cannot ignore.

Artificial intelligence can make scientific discovery faster, but it can also make extraction faster.

Once biological information becomes digital, somebody does not necessarily need the original vial sitting in front of them to continue learning from it.

Information can travel.

It can be copied.

Combined.

Compared.

Modeled.

Analyzed alongside millions of other pieces of information.

And someday it may contribute to discoveries nobody imagined when the information was originally shared.

That does not mean we should stop sharing.

I want to be very clear about that.

A virus does not care about the line drawn on a map between one country and another.

HIV does not ask for citizenship.

Cancer does not stop at customs.

The next pandemic will not wait while governments negotiate who is allowed to open a database.

Science needs collaboration.

Information needs to move.

WHO's current PABS negotiations themselves are wrestling with pathogen materials and their sequence information because modern scientific cooperation increasingly involves both the physical biological material and the information derived from it. World Health Organization

The question is not whether information should move.

The question is:

Can fairness move with it?

That is where I think artificial intelligence could become part of the solution instead of merely another problem we are afraid of.

What If We Used AI to Remember?

What if AI helped maintain the history of scientific contribution while discoveries were happening instead of asking us to reconstruct that history years later?

What if, when information contributed to an important discovery, we could trace the knowledge backward?

Not perfectly.

Science is too interconnected for perfect ownership maps.

But enough to understand the story.

Enough that a country contributing valuable pathogen information does not simply disappear once the information enters an international database.

Enough that researchers from smaller institutions do not vanish from the history of something they helped make possible.

Enough that communities understand how their participation contributed to knowledge.

Enough that when governments begin negotiating access, they know where the scientific value originated.

Enough that when a company says developing a therapy required enormous investment, which may be completely true, we can hold that truth beside another truth: the company did not invent every piece of knowledge that made the therapy possible.

AI does not decide what fairness means.

I would never want it to.

That is our responsibility.

Artificial intelligence does not become morally wise simply because it can process more information than we can.

Human beings still have to decide which contributions matter.

Human beings still have to write laws.

Human beings still have to negotiate agreements.

Human beings still have to decide what level of profit is reasonable, how intellectual property should operate, what governments owe their populations and what wealthy countries owe a global system from which they also benefit.

But AI can help us see the trail before we make those decisions.

That may be one of the most useful things it can give us.

Not an answer.

A better memory.

Now Bring That Back to HIV

This is where that image I saw while scrolling becomes something much larger than an interesting scientific story.

Imagine the AST research continues.

Imagine scientists eventually determine how to use this mechanism safely inside the human body.

Imagine they figure out how to reach HIV reservoirs throughout the body.

Imagine that someday there is a treatment capable of putting HIV into such deep, durable latency that a person no longer needs continuous antiretroviral therapy.

We are not there today. The current research does not establish that outcome. PubMed

But imagine it.

Think about what that would mean for someone who has taken medication every day for years.

Think about the people who remember when an HIV diagnosis sounded completely different than it does today.

Think about the people who watched friends die before effective treatment existed.

Think about the generations of people living with HIV who participated in trials, donated samples, answered questions, allowed researchers into intimate parts of their lives and helped science understand a virus the world once knew frighteningly little about.

If a functional cure comes, it will not suddenly appear out of nowhere.

It will stand on all of that.

And that is why I think we have to start talking about access before the breakthrough arrives.

Usually we do it backward.

We discover something.

We celebrate.

A company develops it.

Patents are filed.

A price is announced.

Governments begin negotiating.

Activists begin asking why people cannot get it.

Countries start discussing manufacturing.

Someone realizes that the treatment requires infrastructure that does not exist in many of the places carrying the greatest burden of disease.

Then we spend years trying to repair an access problem that was built into the system from the beginning.

Why?

Why do we keep waiting until the medicine exists to ask who will receive it?

AI Could Help Us Think About the Ending Before We Get There

This may be one of the least discussed possibilities surrounding artificial intelligence.

If AI can help us understand molecules, why can we not also use it to understand access?

Before a new therapy is approved, we can model where it could realistically be delivered.

We can examine manufacturing capacity.

We can identify where infrastructure is missing.

We can model different licensing structures.

We can ask what happens to price when manufacturing is concentrated in one place versus several regions.

We can identify countries capable of producing components if technology transfer occurs.

We can look at transportation.

Supply chains.

Health systems.

Training.

Accessibility.

We can identify the communities most likely to be left behind before they are actually left behind.

That is AI being used to better humanity too.

Not replacing a scientist.

Not replacing a doctor.

Not replacing an elected government.

Not deciding which patient deserves treatment.

Helping human beings see the consequences of our choices earlier.

WHO itself is already talking about AI and digital systems in contemporary global-health preparedness, including AI-assisted epidemic intelligence, while other international health discussions are emphasizing digital health, artificial intelligence and geographically broader production of health products. World Health Organization

But I think we can go farther.

The question should not only be, How can AI help us discover the medicine?

It should also be, How can AI help us build a world capable of sharing what we discover?

Those are very different questions.

We desperately need both.

We Remember the Patent Better Than the Person

By the time a discovery becomes a medicine, everybody knows the name of the company that makes it.

We can find the patent.

We can find the investors.

We can calculate the market value.

We can usually follow the money almost down to the penny.

What becomes much harder to see are the people and places at the beginning of the story.

The person who gave blood.

The community that participated in the study.

The laboratory that shared a sequence.

The researcher who spent years following something that did not yet look valuable.

The public university.

The government grant.

The clinician.

The nurse.

The activist who spent years forcing institutions to pay attention to a disease they preferred not to discuss.

Somewhere between discovery and commercialization, those human beginnings can become remarkably easy to forget.

That may be where AI gives us an unexpected opportunity.

Machines are becoming exceptionally good at keeping records of relationships.

So maybe the question is not whether AI will become more human.

Maybe the question is whether human beings will use AI to become better at being human.

Better at remembering.

Better at connecting.

Better at seeing who is missing.

Better at recognizing consequences before they become crises.

Better at understanding that scientific value does not begin when somebody files a patent.

And maybe better at refusing to accept a future in which the people whose lives helped create a medical breakthrough are the last people able to receive it.

The Virus Was Carrying Its Own Quiet Code

There is something almost poetic about where this story started.

Scientists have spent decades trying to understand HIV.

And inside the virus itself sits an RNA molecule that appears capable of helping suppress its own transcription.

The virus was carrying information about its own silence.

Human beings had to learn how to read it.

Now we are building machines capable of helping us read biology at a scale previous generations could barely imagine.

We should be excited about that.

We should also be careful.

But careful does not have to mean afraid of everything.

There is a difference between putting guardrails around technology and convincing ourselves that technology is inherently the enemy.

AI on its own does not know what kind of world we want.

We do.

That means the more powerful our tools become, the more important the human questions become.

Who benefits?

Who participates?

Who is remembered?

Who is left behind?

Who gets access?

Who gets to build the next thing?

Who is treated as a partner in knowledge rather than simply a source of information?

Those are not questions artificial intelligence will answer for us.

But it may help make it much harder for us to pretend we cannot see them.

The PABS negotiations are still trying to determine how the international community can share pathogen information quickly while sharing resulting benefits more equitably. The negotiations have not reached their final answer. World Health Organization

The HIV research has not reached its final answer either.

Maybe that is exactly why these stories belong beside one another now.

We are standing in that unusual space between what we can already see becoming possible and what we have not yet decided to do about it.

And this time, we have an opportunity.

We do not have to wait until a discovery becomes valuable before we start talking about fairness.

We do not have to wait until a therapy exists before we ask how somebody in another part of the world will receive it.

We do not have to wait until artificial intelligence has transformed science before deciding whether it should also help preserve the history of who made that science possible.

And we do not have to choose between innovation and humanity.

Maybe the most important thing we can ask of this new technology is not that it think like us.

Maybe it is that we use it to notice the things we have spent generations overlooking.

The connections.

The origins.

The consequences.

The communities.

The people.

And the memory.

Because if the machines can remember, and we finally have the ability to follow knowledge from its human beginning all the way to its commercial end, then the next question becomes much harder to avoid:

What excuse will we have for forgetting?