I am Not Overly Worried About AI Because Not Everything is Computable

There is a good bit of fear among a wide range of people from the everyday man on the street, to politicians, to the media elite, that modern computing (AI) has become or will soon be too capable. In other words, that it can do too many things. These capabilities are being used by some people to perform certain tasks that they formerly would have done for themselves. This includes creative tasks formerly thought to be the sole province of human beings, like creative writing. For the technological utopianists among us it must look like the beginnings of the singularity. Once reached, the human race will be completely freed of the burden of work, living a life of leisure in an ideal world run by intelligent machines fulfilling our every need and catering to our every desire. However to many, many others it looks like the beginnings of the end for mankind. Soon intelligent machines will be capable of performing almost any job a human does, including primarily creative ones, at which point we will no longer be needed. The machines may take pity on us and keep us around for a while, but eventually they will see no need, at which point cue the Terminator storyline, activate Skynet and kiss humanity goodbye.

I have heavily criticized technological utopianism over the years, however, in my own thinking about this topic, surprisingly, I find myself more aligned with their optimistic view, and over time my worry has decreased. Interestingly it feels as if my own worries about this are moving in inverse proportion to everyone else’s with mine decreasing as rapidly as a huge swath of other very smart people’s are increasing. To understand why I feel as I do, it is important to recognize that as a research scientist, I have relied on computers to do a very large chunk of the heavy lifting of my work for almost my entire twenty plus year career. I am not unique in this, nor are scientists as a whole. Many people in many professions can say the same and I think we would all agree that twenty years ago in our jobs and everyday lives computers were absolutely required and/or very, very good for some tasks, somewhat useful but neither particularly good nor bad at others, and terrible or totally useless for others. Today they are exactly the same. The number of tasks at which they are very, very good has greatly increased and they have become even better at many, but the number for which they are totally useless has barely budged, and there still are a lot of these. Even if you disagree, the fact that modern computers can perform more tasks than previous generations of computers has not changed from day 1 of computing until today. Each generation has become more and more capable. You may say, but previous generations of computers could not be used for creative tasks like writing. Yes, and twenty years ago I could have said, previous generations of computers could not be used for creative tasks, like digital photography. There will always be things that modern computers can do that older ones could not, that is the nature of computing which grows in power and speed by Moore’s law.

There is one other thing which has not changed from day 1 of computing until today, and that is the fact that computers can only perform tasks which are reducible to computations (i.e. math/statistics). In other words, that computers today are still computers. It seems stupid and obvious to say it like that, but truly, many, many very smart people constantly forget this or overlook it or ignore it or minimize it. But, it is a brute fact, and ignorance of brute fact does not make it go away. Because computers can only do things which are ultimately reducible to computations their full range of possible future capabilities completely hinges on your particular views on the nature of the universe. If you think everything in the universe (this includes both the biotic and abiotic components of it, including ourselves) can ultimately be reduced to mathematics/statistics than computers will eventually be able to perform any task and do every thing. This includes things like having consciousness, feeling, understanding, having knowledge, having beliefs, etc. If however, you believe, as I do, that not everything can be reduced to mathematics, then there will always be some subset of activities/tasks which computers will not be able to participate in/perform. Thinking that all things can be reducible to mathematics leads to some intractable logical problems. For example, if that were the case it would mean that we are all already computers (or a form of computers). It would mean that all of life and all of the universe is nothing but a sophisticated computer/computer program. In other words, it would lead us right to the simulation hypothesis which I reject because of the too good to be true problem, and other logical problems associated with it. It would also mean that we are no different than computers, that life, is no different from not life. That the abiotic and the biotic are essentially, at base, the same thing. I reject that view as strongly as possible and I find it offensive and even a little frightening. This is because such a view ultimately leads down a path of dehumanization and devaluing of life that I cannot abide and cannot agree with. Do not misunderstand me, this is no religious objection. I have no particular religious beliefs with which I could or would object. This is a matter of self preservation and continuation of the species. Simply put, If we want humanity to continue into the far future, it is essential that we maintain our uniqueness as living things and continue to emphasize the differences between what is life and what is not life. Machines will continue to change and become more powerful, but they will never be alive, because if the day ever comes when they become alive, then they will no longer be machines.