If we can, then sooooo soon

When will we be able to upload our brain to a computer?

virtual_explorer, FirstVDS blog on Habra

"Throne", "Ghost in Armor", "The Thirteenth Floor", "Avatar" by James Cameron, a good third of the episodes of Star Trek… Our public consciousness has long been ready for the fact that our brain can (and probably should?) it will digitize. But how far do we have to go before the technology of full-fledged loading of consciousness? And what chance do we have to see this in the foreseeable future, at least for Musk and Gates?

We often think that human consciousness is the connection of the input—output system of electrical signals with a developed network of processor blocks (our brain). So digitizing all this does not seem to be something impossible. But the reality is more complicated. There are a lot of unknown variables. For example, we do not yet know even the simplest: how much information our brain can actually store.

Different figures are given, most often called 1 petabyte (million gigabytes). In principle, there are already computers in the world that could record all the information of such a brain. But if you delve into this topic, it turns out that a petabyte is just a random number, convenient for memorizing and quoting. In fact, two years ago, a team from the Allen Institute of Brain Sciences in Seattle tried to study at least the structure of the mouse brain. They made a three-dimensional map of all the neurons and connections in one cubic millimeter of space. It took them several months of continuous work to do this. This is considered an outstanding achievement of science. Here is this map:

In this tiny cubic millimeter of brain tissue (smaller than a grain of sand), scientists have counted 100,000 neurons and more than a billion connections between them. And, importantly, they managed to record all this configuration on the computer. Including the shape of each connection and neuron. To do this, they needed... two petabytes of memory. For comparison, all satellite images of the Earth collected by Landsat missions over 30 years take up only 1.3 petabytes.

If scientists now tried to record an entire mouse brain with about 14 million neurons on a computer, they would need 280 petabytes. That's about as much as the whole of Google processes in a day. To store such a brain on the cloud at AWS commercial prices ($0.021 per GB), you would have to pay $5.88 million per month. Do you want to wait 30 years until the technology grows up and you can resurrect this brain? Kindly put out $2.1 billion.

A little expensive, frankly. But what you won't do for the sake of your beloved pet! The brain of which, by the way, the number of neurons is 6,100 times smaller than that of a human.

In general, even if Bezos, Musk and Zuckerberg got together and played roulette to preserve the complete digitized brain of at least one of them, at the current level of technology, their joint capital would be enough for 11 months of storing such a brain (at a cost of $36 billion per month). But in reality, of course, there would simply be nowhere to keep such a volume of data now.

Storage issue

So, to store one human brain, we now need ~170,800 petabytes of data. This is 1.7 zettabytes. It is unrealistically much, but in principle, if we have the task of preserving one brain for generations… Look: in 2020, 64.2 zettabytes of data were created in the world. Hurray! That's 37 brains! Every year!

...But less than 2% of them managed to be permanently preserved. Most of this data was temporarily created or replicated and then deleted to free up space. And 2% of 64.2 ZB is... 1,284 ZB of new data that humanity has been able to save in a year. Even for one brain, such volumes of "new storage" will not be enough.

So we will have to delete some of the information that the Internet now consists of, or require people to reserve space on their hard drives, create smaller sites and stop uploading cats to YouTube so that we, as a whole, can save at least one brain for generations in a year (now the entire Internet, according to estimates, weighs about 40 ZB).

It may seem that this is somehow too much, and this cannot be. Where is one brain that God willing has read a hundred books, and where is the entire amount of information accumulated by mankind. But the number of neural connections in the brain (~2·10 ¹⁴) is simply too large, thousands of times more than the stars in the galaxy. Each neuron is connected to hundreds or thousands of others, and the complexity of the resulting structure is incredible.

Of course, part of the reason here is the imperfection of our current brain scanning technologies. If you take the case from the laboratories to the commercial market, there will probably be companies that can compress the data 10 or 100 times. And they will certainly offer us the service of removing the cerebellum — after all, why do we need motor skills if we live in a cloud. Occupying only 10% of the brain volume, the cerebellum contains 77% of our neurons. And this is 77% of the potential savings!

And when they finally invent androids in which our brains will be transplanted, we will copy the cerebellum from someone else. Well, the gait will change a little, it's okay. But we will learn to tap dance! And we will be able, for example, to order a copy of the cerebellum from some Asian in order to finally use sushi sticks perfectly. And copies from the best dancers and ballet dancers will be in great demand.

Oh, let's live!

Bandwidth issue

But data extraction and storage is not the only problem. In order for a computer to work in brain mode, it needs to be able to quickly access all this stored information. In other words, a significant part of this data must be stored in RAM. Some kind of cloud storage is not enough here. There should be a short neural connection between each first and second bit, and not thousands of kilometers of wires leading to another data center.

And here is a problem of a different order: the largest computer ever created with a single memory module had 160 TB of RAM. This is about a million times less than the 1.7 ZB that we need. Even if Moore's law works by some miracle for brain storage technologies (which has exhausted itself in terms of improving data storage facilities for several years), it will take us at least 40 years to reach the required capacity. That is, if we throw all our efforts into development right now and do this business the same way we did reducing the size of transistors, by 2062 humanity will have a working model of the brain on a computer capable of reacting at the same speed as our current brain. This is under the best circumstances, of course.

By the way, here we forgot to take into account that any error when transferring data can be fatal. But we do not know how the information is organized, where the most important details for this particular person are. We'll have to duplicate the brain several times and then compare copies with each other, figuring out which one turned out to be more accurate. And then, of course, store all this data in at least two (if not three) copies to prevent their loss. Which, unfortunately, can happen on a computer at any time, for example, from the same cosmic rays knocking out bits.

Information in the brain is stored in every detail of its physical structure, in every connection between neurons, in the nature of each neuron, in its size and shape, chemical structure, as well as in the number and location of connections. All this will need to be perfectly copied, and then contained in several samples. In general, RAM manufacturers will be happy.

A matter of time

After death, our brain changes very quickly — both chemically and structurally. When neurons die, they quickly lose the ability to communicate with each other and cease to show their characteristic properties. But even more problematic is the fact that our brains change a lot during our lifetime.

Since the age of 20, we have been losing 85,000 neurons a day. This is not a reason to worry: mostly those neurons that have not found a use for themselves die, they were not involved in any information processing. If their inactivity occurred for a long time, this triggers a self-destruction program (apoptosis, a regulated process of cell death). In general, several tens of thousands of our neurons kill themselves every day.

But this is not too big a problem, because in 20 years we have 90-100 billion neurons. And at this rate, we will lose only 2-3% of our neurons by the age of 80. If we don't get a neurodegenerative disease, our brain can work well throughout our lives. But then the question arises: at what age is it better to stop and do auto-save?

Which mind would you prefer to keep, yourself 20-year-old or 70-year-old? In the first case, the brain will be more plastic, but in the second case, you will have already accumulated knowledge. In addition, you can save 2-3% on the cost of scanning and data storage. But if it persists too late, you can get a digital brain with dementia that has begun to develop. Or a brain with the beginnings of Alzheimer's disease, which can begin to develop even before the age of 40.

So what to do? At what age are you the best you are? When do I need to save myself? Is it worth sacrificing all your experiences and memories? If you decided to keep yourself 30 years old, and died in 80, apparently, you need to write down a notebook for you with a list of all those events that happened to you over these 50 years? And a couple of photos from the family album to show how it all ended for you? 

The question of ethics

What to do with the time that you have lived, but you will never remember? Won't your digital self always feel sorry for the irretrievably missed days?

And even if it doesn't, how will you know that it's really you? How do you check that you would make exactly the same decisions, that you have a common train of thought? At least when your brain was preserved, your real self went on a completely different path, it continued to develop for many more years. Your computer self has little in common with him anymore.

Many would say that your mental image transferred to a computer is no more alive than the computer on which it is placed. It's just a pirated copy of a living person, without a license and half of the features.

The living mind draws information about the world through the senses. It is attached to a body that exists in the physical world. Who has a changing heart rate, breathing and sweating, who feels hunger and fear, whose mood depends on how well he slept and what kind of dream he saw. How will all this work in a computer without a body, on bits and bytes?

But we also need to take into account the physicochemical nature of the brain itself. Neurons are not always the same, there are different types, plus they change periodically, as does the nature of the connections between them. Even in adults, as science found out in the early 2000s, neurogenesis occurs (the fact that "nerve cells are not replenished" is a myth). All these complex biological and chemical processes will also need to be simulated somehow, having only zeros and ones at hand.

Even just modeling the "physical" input and output is already a big problem. We will need cameras that can accurately copy the human eye, skin that can respond to touch, nostrils that can detect aromas. There's nothing to say about internal organs, it's even more complicated. If we could do all this, we would have a ready-made person.

To grow a new person in a test tube and "transplant" your brain to him, having pumped the data back, will not work either. Even if you're a billionaire and you don't care about ethics. There is simply no other brain with exactly the same structure, size, and all chemical and biological features. The best you can count on is that you will live forever in the body of an android. With imperfect senses. Every day, driving away from yourself the thought that you are not living, but only imitating real life.

In other words, with all the sacrifices and billions spent, we can never be guaranteed to keep our brain alive. Simply because we won't be sure if this new state is life. Your transferred mind will not be able to relate to the world in the same way as your current living mind does. That is, it won't be quite you. This is not to mention the danger of hacking (and what if you were hacked while you were in "hibernation"?) and hardware failure.

Is it worth spending billions to preserve the brain structure of someone who won't be quite you? And guaranteed to never be able to live fully?

Or is it better to spend this money and time on solving other pressing (and, most importantly, really solvable) problems?

Here it remains only to wish all Habrov residents a long and healthy life) Because we definitely have one. Even if the elements of some of us in 40-50 years will begin to exist on the AWS cloud or in the form of androids.

Portal "Eternal youth" http://vechnayamolodost.ru