Will machines think like human beings, or will human beings begin thinking like machines?

This is an interesting question with rather practical implications to a programmer. Although there are certain advantages understanding the low level mechanics of how computers work, I don't think the relationship between that kind of technical knowledge and being able to write a program are necessarily related. No doubt a relationship exists, but I don't think it follows that knowing one automatically leads to the other.

Perhaps an analogy would better explain what I'm getting at. Technical details are like the study of linguistics compared to programs which are like prose/poetry. Knowing linguistics allows you to describe languages and their components such as verbs, nouns, adverbs, the rules of grammatical correctness, etc. This knowledge however doesn't mean that familiarity with these structural elements will automatically allow you to write meaningfully or expressively. These elements provide the constraints of expression when attempting to articulate an idea, but the articulation of the idea lies outside the bounds of this domain of knowledge. This is the same in programming. Although understanding how a computer functions or a compiler works is useful, it does not follow that you can write useful or meaningful programs just as knowing linguistics won't necessarily let you write literature. So what exactly is the nature of the relationship?

At its core, I think that the relationship can be summed up as follows:

Programming, as an endeavor, is still an exercise in expression.

How you choose to express an idea is as important as the idea itself. Like great literature which is firmly rooted in the language chosen by the author, a program acquires this quality of expressiveness based on the language which forms the foundation of a program. Although great writers sometime "break the rules" when writing literature, it is no different than the clever hacks exploited by programmers to express the difference between constraint and convention.

This idea of language as a source the expression implies that certain languages are better suited to expressing different ideas. No one single language is capable of the full range of expression because, at the point of its creation, the range of future expression is completely unknown. It also follows that languages evolve to fill that gap in expression.

So what is it that programmers seek in languages that they choose to solve problems? Quite simply, languages with real longevity possess an expressive quality that is both aesthetic and technical. Although semantics may vary widely between languages, this "expressiveness" is what programmers implicitly recognize and use to build their conceptual framework on which to hang those semantics. For lack of a better word (and borrowing shamelessly from Neal Stephanson and his novel Snowcrash), this is the Ur quality that is the root of all programming languages.

This brings us back to the question at the beginning of this article. Implicitly, expression is still a very human endeavour. Until this changes, I'm inclined to believe that machines will evolve to express very human ideas which will be reflected in the language we choose to express our intentions to them.

I think in this regard, Ruby is a step in the right direction.

Addendum

I found this article a day after writing this tidbit. Interesting how much the aesthetic guides intuition to the truth.