The New Therapy

 "But why does the idea make so upset?” she asked.

“Because now the people who are alive are not free, and to achieve this trillions of other people die every day. I can’t even understand the magnitude of this tragedy, and it’s my fault,” I explained.

“Hmmm, please tell me this story the way you see it”

“Okay:

When I started studying psychology I never thought coding would be such a big part of it, but in my defense, at the time it wasn’t. As my studies continued the ability to simulate human minds became better and better, and this happened extremely quickly.

I remember my professors discussing these new technologies in class with us, but without taking it too seriously. In Taiwan someone had mapped out memories in the brain, in Oman someone had stimulated neurons to make people hear musical notes, in Nigeria someone had connected a brain to a computer… they were just curiosities.

But then in my third or maybe fourth year we had a project about depression. We formed teams and each team got a volunteer, someone with long term depression. We had to help this patient using some of the techniques we had learned, and find ways to measure their progress. I remember a guy, one of those super serious super smart guys who don’t make any friends, you know the type. He brought his computer and he had software that supposedly simulated the overall mental state of his team’s patient.

It didn’t work really well, we got the same grade at the end, but looking back the fact that he could even have a rough computer simulation of what his patient was feeling… it was a prophecy of the times to come.

When I graduated I decided to continue with a masters degree and I decided to go deeper into the clinical use of human simulations. My thesis was about how we could even distinguish a depressed mental state from a regular mental state in a simulation. I researched what objective properties of the simulation we could use to figure out if this was a simulation of a depressed person or not.

After that I got a job at a relatively new company. They wanted to create software that psychologists could use to give better therapy to their patients.

Initially it was so harmless. This is how it would work:

A psychologist would greet a new patient and ask them a few questions. Unbeknownst to the patient the room had many devices, such as microphones and cameras, that would gather a ton of data about them in a very short time, including their answers to the questions.

Then our software would use all this information to create a very simple simulation… but it wasn’t a simulation of that person. You couldn’t talk to it, it couldn’t answer questions, it didn’t have any of the memories of the patient… we called them “simos" for “simulated overall state.” This virtual creature was just an approximation of a few properties of the mental state of the patient at that time.

For example, once the simo was created our software would inform the psychologist about the emotions of their patients with an insight and precision they could not get any other way. Then as the therapy session progressed the simo would be updated in real time. Using this he psychologist would be able to tell exactly how their questions and suggestions had affected the patient in real time. Among multiple sessions they could use statistics to quantify which strategies worked best and use that to plan better therapies for their patients.

If it had been only that, it would have been great, we would have certainly avoided a ton of problems… but when have humans ever just stayed satisfied?

I guess part of the problem was the personhood status of the simulations.

At the same time we were selling this software other researchers in universities and technology companies were trying to take human simulations to another level. They dreamed with being able to turn humans into purely simulated beings, unshackled by the limitations of our physical bodies. To this end they created real simulated people. Beings you could talk to, beings that had the same memories a person would have, beings that could feel emotions and discuss philosophy and be bored by philosophy and want to watch a movie instead. 

The process to achieve this was quite difficult. They had to take the brain of a deceased person, then they had to take all the nerves (like the eyes, ears and the spinal cord) and very carefully connect those nerves to their machines. After that they had to surround the brain with an army of sensors.

Once the setup was done, they would start sending signals to that brain through the nerves and the sensors would measure how that brain reacted. This process would continue for months, always collecting data and it would only stop because the techniques used to avoid the rotting of the brain had reached their limit.

Finally they would give half this data to a computer program and the program would generate a virtual brain. Then they would send signals to this virtual brain, signals identical to the other half of the data. Amazingly, the simulated brain would react to those signals in the exact same way the real brain had reacted. Meaning it had to be a faithful reproduction of that real brain.

This was so effective that if you allowed this virtual brain to speak they would not just pass the Turing test, they could fool the friends and family of the dead person into thinking it was the real person. Of course it would not take long for the simulated brain to remember how the real person had died and start freaking out, at which point you would just reset the simulation to the starting point.

A few videos of virtual brains freaking out became public and many people claimed they were real people. Everyone thought that except the people actually working to create those simulations.

I guess, when you make something, and you understand deeply how it works… When you see it working, you only see gears turning, you only focus on the imperfections and get ideas on how to improve it… 

Don’t get me wrong, scientists absolutely get attached to their inventions, but for these people their invention wasn’t those simulated people, but rather the system itself that allowed them to simulate people in the first place.

After they had a few of these complete simulations the researchers wanted to study more simulated minds. The problem was how expensive and time consuming it was to digitize a single mind. So they figured they didn’t have to digitize more real people, they just had to simulate random realistic people.

Here’s what I mean: they figured out how to simulate a new random person with just the data they already had. With just a tap on a screen their program would start mixing its database of simulated brains to create a new simulated brain. 

It was like taking photos of many human faces and then cutting the eyes of one photo, the ears of another, the mouth of a third photo and so on, and then putting them all together to make a new face, one that didn’t appear in any of the photos, but a face that looked realistic nonetheless.

This is how they were able to create simulated brains so realistic the could very well have been simulations of real people. Then they could tap on the screen again, that simulation would be stopped and another simulated brain would be created.

As their research continued they would simulate and stop simulating thousands of people per day in tens of research centers around the world. 

Some said this was genocide, but the scientists argued most people just didn’t understand.

And I have to admit, I agreed with them. This was just software. These minds weren’t any more real than a tamagochi or any kind of virtual pet. Sure you can get attached to them, but you have to keep in mind they are just imitations of the real thing, and no matter how good the imitation gets, it is still just an imitation.

This was at the core of what the universities and technology companies argued in front of the Human Rights court of the UN, and they won. 

For that reason I was among the first people in my company who campaigned to make a deal with one of these universities and have a chance to incorporate their new technology into our product.

At first the problems seemed unsurmountable. Taking enough data about the patient to reproduce their mind faithfully was impossible unless we killed them. This wasn’t an option for us, obviously. We could only have these small snapshots of the mental state of the patients.

But I solved it. I found the solution. It was my masterpiece.

Remember how it was possible to create a realistic random simulated brain?

Well… what if we did that millions and billions and trillions of times?. Then we could take snapshots of each one of those brains until we found one who in that moment, with the same stimulus as the real patient, produced an identical snapshot as the real patient.

Do you get it?, we ended up with two minds that produced the same snapshot, that’s basically impossible, that just doesn’t happen… unless… unless both minds are the same!

It was like cutting faces from magazines and mixing their attributes until by pure chance it happened to look like our patient.

Of course, in practice it was a lot more complex. I’m giving you the simple version, the one I used with my husband, he’s not into all this sort of stuff, but he understood that explanation, so I figure if he understands then anyone does.

Anyway. In practice it took several sessions, we had to take many snapshots and use them to  guide the brain simulating algorithm, making it faster and more precise. It took hundreds of people working for nearly a year, but in the end, it worked.

Psychologists could now have full simulations of their patients and use those simulation to help those patients in ways that would have been basically impossible before.

Of course the product still wasn’t sold commercially. These were just our own experiments with volunteers and our own team of psychologists. In fact this product was never sold, because soon we came up with something even better.

The idea was so simple, it happened so innocently.

We were talking with one of the psychologists about her experience using the new prototype, when she said:

“I wish I could ask this simulation a few things about the childhood of my patient. He’s not opening up about it.”

We looked at each other, and we thought we could make it happen.

We didn’t talk to the simulations directly. That’s terrible idea. They are very confused and if they understand what’s going on they start claiming they are real and beg you to not turn them off. It’s a bad experience.

Instead we often keep the minds “frozen” until we need to stimulate them, at which point we simulate a virtual environment for them to interact with in ways we can control.

What if we added a fake psychologist to this simulated environment?, and that psychologist could ask the simulation about their childhood and this data would be very useful to help the real patient.

It took us like three months to set it up, but we did. 

During development we had an obvious question: if the real patient doesn’t open up, why would the simulation?, it is supposed to act like the real person after all. The solution was clear: we just try again. We reset the simulation to a previous point and we repeat the session over and over with slight variations each time, until we find out what works.

The first few test were a success.

Psychologists could make simulations of questions or situations they wanted to test out and then be able to repeat those experiences in real life, and it worked, they achieved the progress they were looking for, and more. 

This whole endeavor had taken months and millions more in research, but it was infinitely better than the product we had before… until there were problems, one big problem actually:

Sometimes it was really hard to repeat what had happened in the simulation.

It makes sense. The patient leaves, lives their life, things happen that you don’t expect, and those experiences make your patient a different version of themselves than the one in the simulation. This means they may react in ways the simulation didn’t predict, making the simulation itself useless.

Sure, the simulation can be updated, but the point was to be ready before the patient arrived. You need to be ready to help the patient you have now, not the one you had a month ago.

This was the last domino, the last of our bad decisions in a chain of bad decisions. Looking back I wonder if there was anything we could have even done. We had been pushed so far down this road, perhaps we could not have turned back even if we had wanted to. But the point is that we didn’t want to turn back, we thought we were making great decisions.

We heard about another development. The governments of the Triple Alliance and the EU had developed a new technology that made it possible to alter the mind of a person… they had developed this technology to the point  you could basically upload any information you wanted into a brain, and even delete information if you had to.

The armies of those nations were using it to create super soldiers or something like that but we got permission to use it for what we saw as a much more benevolent purpose.

How do I even explain it?, I’m ashamed even to admit it…

I guess, it was all because of Laura, one of our volunteers. She was so depressed, she had so many problems. Medication had not worked for her at all, and therapy was not very successful either. We all liked Laura, we all wanted to help her and people like her.

One day one of our technicians was working with a simulated Laura trying to get her to open about a traumatic experience she had, but simulation after simulation nothing worked. In frustration the technician left the program running at maximum speed, which we basically never did.

Billions of Lauras sat through billions of therapy sessions, each one slightly different than the last, trying new things every time.

Then, it worked. The technician couldn’t believe it. The simulated Laura had not just opened about her traumatic experience, she was not depressed anymore.

We had a meeting just to analyze the results. This new simulation was a version of Laura, that clear, but this simulation did not have any indications of being depressed. It was normal, it was just normal.

Somehow the program had found a single therapy session so successful it had cured Laura of all the problems she had struggled with throughout her life.

We of course tried to repeat this miraculous therapy session in real life, I swear to God we tried, but it just couldn’t be done. Laura would rarely react as expected, and the slightest variation in timing and tone of voice form the psychologist would derail the whole thing.

At last I thought: why don’t we upload into Laura the simulated Laura that went through the successful therapy session?.

We talked with Laura about this, we explained everything to her, and she agreed. We did it, and it was beautiful. The Laura we met that day was the person we all knew she could be but had always been trapped under her depression and traumas.

Why not do this every time?

Seriously. We could run simulations until we simulated one therapy session that would cure our patients of all their problems, and then we just had to upload this new version of themselves into their brains and they would be cured. You only had to do it once and it had a 100% rate of success.

This is the product we finally shipped. In a few years, depression had been cured all over the world, along with many other conditions and mental disorders…

But then Saudi Arabia declared that not believing in God was a mental disorder, and people would go through therapy against their will, being replaced with a version of themselves who did believe in God. China declared that not trusting the government was a mental disorder. Some states of the USA declared that not being christian was a mental disorder… it kept going.

We had created a tool that allowed to edit people into any other possible version of themselves.

When I realized this, as I saw the abyss filled with the millions of people who were erased from existence to be replaced… I finally had to accept that I was wrong. If the original versions who were replaced were people, and this replacement was a form of murder, then simulated people were real people too… we had killed… trillions of people with our simulations… we just didn’t care…"

“But how can it be murder if the person is right there?,” she asked “walking around, living their life. No one died.”

I looked at her. Seating on her chair, holding a pen and a notebook, looking at me.

“I am in therapy, aren’t I?”

“Reset" she said.