To some, the answer may be "well, duh." To others, it may seem that I'm making a poetically admirable but superficial statement that's obviated by the fact that everything is a collection of small atomic parts that interact to form an emergent system.
As the title suggests, I'd like to make an argument against the latter; but I'd also like to ask the former group to read along. Why? Because this isn't based on a metaphysical defiance of science but in fact something of a declaration that science and spirituality are quite compatible; in fact, I will be saying that reducing human beings to machines is deeply un-scientific. But I digress.
First thing's first: yes, using a very loose definition of machine* you can argue that all living things are machines; but it's a pretty banal and in fact misleading definition. Why misleading? Because the word machine has a whole lot of connotations and when it comes to a word, you can change the definition without sufficiently changing the connotations of the word. Machines are seen generally as artificial, unthinking, heartless, cold and calculating (I could go on, but you get the point.) But could we really make these accusations at a phenomena so general and abstract as a system that emerges from interacting parts?
These misleading connotations have a particular moral hazard; they encourage a very mechanistic and quite possibly nihilist view of the world. I'd like to note that I'm not saying that a mechanistic interpretation existence is bad in itself; that's science and I happen to think like most people that science is a very good thing. But when we take the word machine, with all its implications, and slap it on everything we see regardless of the context in which we're talking about it, our world view starts to become an outright perversion of science. This isn't just some spiritual quibble; in many ways it leads our thinking to become deeply un-scientific. This may sound odd, but in order to make this point I need to touch upon the cultural properties of machines.
Machines were historically created as something to aid and simplify human labor, starting as tools in the hunter-gatherer period. Since then, we've gotten to the point of automated assembly lines and computers. When making a machine, one generally needs to specify the problem and the solution in a way that can easily be understood as separate parts; not only in order to come up with a feasible design before building it, but also for the sake of being able to figure out what went wrong if the machine fails. For the entirety of Civilization, machines have been understood as a configuration of parts that can be understood how each part contributes to fulfilling the machine's function and whose actions are predictable. This sounds like a loaded statement, but it only comes from the logic of why and how we create tools and machines; by the fact that it was necessary to produce a reliable outcome and that in order to do so the logic had to be sufficiently simple. Note that this even applies to machines like random number generators; we may not be able to predict the number, but we can fully understand and predict how each part will contribute to the process of creating that random number.
But humans are hardly like this (or any animal for that matter.) We may be able to derive general ideas about living things by discovering the most basic moving parts or performing some specific experiments, but the interactions between these parts and the emergent patterns that come from them are well beyond our current comprehension. Unlike machines, organisms are almost entirely unpredictable. In the paradigm of physics, this doesn't matter because physics is only interested in the simplest forces and smallest parts; no physicist has to actually come up with any predictions about the human condition. In this context, it's perfectly safe to define a machine as anything that converts energy from one form to another.
But what of the many other things that we're looking to understand? By taking this definition of machine from physics and glibly applying it to every other schema through which we look at humans, we've turned a blind eye to the fundamental randomness of organisms. By this, I am not making any metaphysical argument about free-will/chance/determinism, but using randomness in the mathematical sense, which simply states that if you can't predict it (due to a lack of information or otherwise), then it's random for all intents and purposes. It's no wonder then that we've failed to create AI that portrays humans in any life-like manner, that we still wonder why shamanism or religion is sometimes a better alternative to clinical treatment** or that a thousand economists with PhDs couldn't see the imminent collapse of the world financial system.
The mechanistic view of the world is appropriate for fields and paradigms in which its objects of study can be sufficiently understood in such a light. Once we let this view pervade the rest of our thoughts, we end up looking at the map and not the territory. Calling living things "machines" in a general sense does just this by implying that their parts and behaviors are comprehensible as such and gives the world a false sense of predictability. Reducing the complexities of life in this fashion isn't just offensive, it's moronic.
*The formal definition of machine is an object that converts energy from one form to another. By this definition, anything that materially exists is a machine.
**For the record, I am not trashing clinical treatment of people in need of help. I am the son of a psychologist and a pediatric nurse. I believe that medical professionals are oftentimes helpful because they are usually very scientific in how they rely on data. I should also note that my complaint with the mechanistic view of the universe is not with empiricism; true empiricism acknowledges what we don't know and doesn't rely on representations. What I am suggesting is that in the face of randomness, clinical treatment doesn't always make sense.