The Kurzweil Singularity

On Friday I attended a speech by Ray Kurzweil at the 2nd Interdisciplinary Graduate Conference on Consciousness. The speech was, first and foremost, a defense of Kurzweil’s prediction that by the year 2029 technology will be advanced enough to combat the process of aging by modifying our bodies and even our brains so that we can live forever. This obviously raises a number of existential questions as it will drastically change our social dynamic. However, Kurzweil focused on defending his prediction and confronting the dilemmas that arise in our understanding of consciousness.

In support of his thesis, Kurzweil predicts that machines will have human capabilities (in terms of consciousness) because of the exponential growth of technology. The remarkable thing about its expansion is that it is completely predictable. Kurzweil has spent time tracking the advancement of technology by the unit Bits-Per-Dollar, which I interpret to mean the ratio of amount of information to its cost. Without getting into the mathematical aspect of his theory, what is important to recognize is that when the ratio is graphed over time, not only is the growth shown to be exponential, but it reveals how incredibly predictable it is.

One important caveat to notice is that these predictions, while they have been accurate, are not predicting specific devices, inventions, advancements, etc. Rather, they are simply predicting the path of the whole power of technology over time. I will bring this point up later.

Kurzweil’s general epistemology states that knowledge is nothing more than a pattern; and technology has the ability to reproduce these patterns, even in the individual. To Kurzweil people are also nothing more than a pattern. For example, the entire “content” of our bodies (i.e. our cells) changes thousands if not millions of times over our life time. But the only continuity among all these changes is the pattern they follow and our consciousness. Hence, our consciousness must be some kind of pattern that, according to Kurzweil, technology will be able to reproduce.

Issues

There were two issues brought up that I wish to discuss. The first is simply an interesting dilemma, the second, an investigation into the phenomenal experience of a machine with human capability.

Kurzweil introduced a dilemma with the following example. Suppose that we made an immediate copy of Ray, such that every tiny aspect of Ray was copied into another (“separate”) individual we will call Ray 2. Ray 2 is everything Ray is, but he can live forever. But we are not about to turn to Ray and tell him that he is now disposable, for Ray and Ray 2 are still separate entities. There has been no transfer of consciousness to Ray 2, though Ray 2 has consciousness which is empirically identical with that of Ray. So, in some odd way, Ray 2 is not Ray. (Doesn’t this present a problem for what I mentioned above? If we replicate the pattern, as we did with Ray 2, it does not end up with the “same” consciousness. So, consciousness is not a pattern that can be replicated, it would seem).

However, if we were to gradually advance Ray piece by piece—first starting with mechanical limbs—and eventually re-wiring him completely so that he is no longer a biological entity, we could then say that we have maintained Ray’s consciousness. However, this gradual Ray (I will call GRay) is identical with Ray 2, who we just said, is not Ray. Even though it seems more reasonable to assume that GRay is really Ray, the problem persists. I believe that this requires an investigation into the nature of phenomenal identity that transcends the scope of this discussion.

Second, there was a question raised about the phenomenal character of a conscious machine. It seems reasonable to assume that there ought to be a distinction made, when constructing a machine with human cognitive abilities, between the capabilities of consciousness and its characteristics. By “characteristics,” I mean a kind of phenomenal “feel,” such as, what it is to experience seeing the color red or getting a headache.

Kurzweil immediately recognized that the issue is dependent on an interpretation of “qualia,” by which I was impressed. However, he failed to really answer the question. He turned to the philosophy of language in order to explain how, even with biological consciousness, qualia is something that cannot be accounted for. We have been conditioned, he argued, to call certain things ‘red’ no matter the way the object really is; i.e. the problem of red-green color blindness. It was not the answer I expected and, to me, it only further illustrated that consciousness is something so totally out of our comprehension that replicating it seems improbable, even in the face of Kurzweil’s support for the predictive element of the growth of technology that I will touch on next.

Objections

I mentioned before that what Kurzweil is doing is predicting the whole power of technology based on the pattern of exponential growth that has revealed itself over time. However, I suspect that his prediction might be inaccurate because of this. Kurzweil even said that the prediction does not account for individual inventions, advancements, etc. It only follows the ratio of bits of information per dollar. So, how can he predict something so specific from something so broad? If he doesn’t know what precise technology is necessary to create mechanical consciousness how can he say when it will happen? Even if you were to give him the benefit and say that he can, in fact, predict the power of technology into the future, how does he know how far technology must be advanced in order to create the Singularity? Though I’m sure Kurzweil has an answer to these problems, they were not addressed on Friday.

0 comments: