Is it possible to “upload” a human mind on a silicone substrate?
A dynamical systems perspective
In response to this article by Trent McConaghy.
In his article, Trent argues that human patterns of intelligence in our brains / nervous systems / bodies are just one instantiation of this dynamical system, and that, once you have a model of that system, you can instantiate those dynamics in whatever substrate you please.
From my experiences with modeling highly complex nonlinear dynamical systems I see some fundamental methodical and physical limitations here that I would like to discuss. My personal object of study is the earths atmosphere, which might seem unrelated at first glance, but in essence is just another nonlinear dynamical system, maybe less complex even than the human mind-body-system, and definitely infinitely better understood. But even there with our simulations we run into some very fundamental barriers.
Let’s assume we had a thorough understanding of all the dynamics and mechanisms involved in the human mind-body-system, similar to where we stand today with the earths atmosphere (quite an assumption in itself already), we then still have the fundamental problems that a) we are still using reductionist models, not simulating the actual processes, and b) we can’t properly observe the initial state of the system down to every detail. In the simulation of highly complex nonlinear dynamical systems both of those problems have profound consequences. As soon as you start using even the slightest simplifications and approximations you are getting into serious trouble and your model will diverge from reality fast, and eventually blow up at some point. To keep it in check you will have to introduce additional rules and boundary conditions. But then you are simulating only something that looks somewhat similar to the original system, but actually has a fundamentally different nature, and little to do with the true original dynamics. In the case of the emulation of human intelligence you probably literally take the humanity/life out of the intelligence through such a fundamental reduction of complexity/depth and randomness. It still might be some kind of intelligence, but nothing resembling human intelligence.
So when confronted with a highly (this word is important here!) complex nonlinear dynamical system you have to choose: either you put in the resources to simulate it in its full complexity, which in the discussed cases probably only would be possible if you would simulate them down to the interactions on a molecular level, or you get a very limited model of reality that actually has a fundamentally different nature and outcomes. Today’s atmospheric models, despite using some of the largest super computers in the world, enormous global observation networks (including weather stations, airplane sensors, RADAR, LIDAR and satellite imaging), and decades of research and development by thousands of highly talented people, still only are able to simulate a few days of the evolution of the earths atmosphere, and in many aspects with a very limited resemblance of the true observed phenomena. Those models only work due to the use of a lot of rough approximations and parameterizations, and only can be kept in check with the help of a lot of additional, artificial rules and boundary conditions, and therefore actually do not simulate the true dynamics of the system and its inherent natural “intelligence” at all. A lot of this has to do with a lack of observations/input data and a lack of computational resources (which also are huge problems with the brain-body-system), but even if you theoretically could observe the exact initial state of each molecule and all the correct boundary conditions as input for an atmospheric model (radiation in/out, boundary interactions with the earths surface and outer space), and if you had the computational capacity to calculate all those interactions between all those molecules (and photons), then still, due to the fundamental nature of chaos (and quantum) dynamics, you only could simulate around a maximum of two weeks into the future. This is a final, insurmountable barrier set by the laws of the universe, similar to the speed of light. And considering the number of neurons in the human brain (and body) and their highly complex interconnections and pattern dynamics, and all the chemical and electrical (and maybe even quantum) mechanisms involved, and the nature of the boundaries to the surrounding world (the rest of the human organism, and even the outer world), I see a similar degree of complexity in the human mind-body-system, and therefore the same fundamental limitations.
From the perspective of modeling highly complex nonlinear dynamic systems there seem to be fundamental methodical and physical barriers to instantiating the dynamical patterns of human mind-body-intelligence onto a silicone substrate. It should be possible to build some kind of model that is able to replicate some aspects of the behavior of the original system, but such a model won’t be very close to the original nature of this system and deviate rather soon and fundamentally from it. It will contain some kind of intelligence, but not human intelligence.
Even if those barriers for some reason in the end shouldn’t turn out to be so fundamental (which I really can’t imagine, since they are based on fundamental laws of physics and chaos theory), the remaining hurdles still seem way too high to overcome in the time left until AI “wakes up”. In this article Trent is pointing out why this could happen much sooner than most of us expect. This seems plausible to me, and therefore I see it as highly unlikely to even get somewhere near to having the technology to scan the brain in sufficient detail and/or to truly understanding its mechanisms and dynamics before AI wakes up.
We should focus our resources on other approaches instead.
Ideally, of course, there should be something like a council of the wisest men and women on the planet, carefully weighing the decisions on what steps should be taken in the development of AGI and what steps should better be left undone. But in the way in which humanity currently is organized (*cough*, *Moloch*, *cough*) I can’t see that happening in the required time span anymore.
And maybe we just should have a little more faith in the nature of the universe/the hypergraph/god. After all, if an AGI emerges that is 100x more intelligent (and wise?) then humans, why not let IT decide if organic life and humans are worth keeping around or not, and in what form? Per definition it should have deeper insights into the nature of the universe than us. And wouldn’t it be plausible that it decides to do so? Just as intelligent, wise humans usually care about other forms of life? (and the more so the higher the intellectual and emotional/spiritual development) … this only refers to a true, far super human AGI though. There still would be the risk of someone building a powerful pre-singularity AI with malicious, manipulative, exploitative intentions (*cough*, *facebook*, *cough*), or a paper clip maximizer with badly constructed incentive systems, or something along those lines. We definitely need to stay alert to such developments and try everything in our hands to intervene. … but even if the experiment of human life on this planet should fail: in the end, that might just be the nature of the tinkering universe... it was a nice try, but maybe humanity just evolved into a dead end, and another experiment of life on this or another planet someday will be more successful.
But in any case, we always need to retain a positive spirit, otherwise we already have lost. And there is reason for this. After all, humanity is in the possession of nuclear weapons since more than 70 years now, and it didn’t blow itself up yet. Environmental destruction in developed nations is in the process of getting reversed. We always seem to be dancing on the brink of the abyss, but in the end to not fall. So no matter what lies ahead, let’s focus on hope, stubborn optimism, and work with wisdom, love, persistence, joy and creativity in the right direction. The world is not lost yet. But it is not saved yet either. Let’s give it our best.