There are no artefacts, at least not in any metaphysically serious way. To see this, consider some puzzles that face those who believe in artefacts.
You come across a stump in the woods. You sit down. Has the stump become a seat? It seems so. Even though you haven't actually changed the stump, you have given it the function of a seat. An artefact does not, after all, need multiple pieces. I could manufacture wooden seats by taking logs and carving them into seats, or by spending a lot of time in the woods and finding pieces of wood that already have the right shape and bringing them back. Or perhaps it's not enough just to sit on a stump to have a seat.
But actually, if I manufacture seats, the stuff I make does not become a seat when someone sits on it—it is a seat when the manufacturing process is complete. Likewise, then, it seems that the stump has already become a seat when I intend to sit on it. Or if giving something the function of being sat on once is not enough to make it a seat, maybe it becomes a seat as soon as I intend to take it home and put it in my living room. But this is really spooky: by simply intending to treat a stump a certain way, I have made a new object—a seat—with new persistence conditions. Very strange, that.
You buy a chair. Dust settles on it. The dust is not part of the chair. But you then make a tag: "Dusty Chair II". You put the tag in front of the chair. Oh, I "forgot" to say—your home is an art gallery. When you put the tag in front of the chair, with the right intentions, you created a work of art, an artefact with certain complex persistence conditions. One of the persistence conditions is that while the dust is not a part of the chair, it is essentially a part of Dusty Chair II. So, simply by putting a tag in front of the chair, I have brought it about that a new artistic artefact comes to exist. The token tag is not, however, a part of the work of art. (The signs in a gallery are printed by the gallery, and need not travel with the works of art.) So I created a new entity, without in any way causally interacting with it—just by putting a tag in front of it. And of course the tag is unnecessary. If I just stand in front of the chair telling all the visitors to my home that this is Dusty Chair II, that's just as good.
The case of the stump-seat and Dusty Chair II violate the principle that one cannot bring about the existence of a new object without a relevant[note 1] causal interaction with the object. Artefacts are simply too easy to make.
The above arguments assumed that what defined the identities of artefacts were maker's intentions. The alternative is social practices. But it is no less weird to suppose that a bunch of people by getting together can create an object without relevantly causally interacting with it.
It is clear that the persistence conditions for artefacts are defined by the makers and/or users and/or the community. But let us say that a careful study of our language shows that half of the users of English understand "chair" as an essentially four-legged artefact, and the other half think of a "chair" as surviving the loss of one of its four legs. Does that mean that my dining room chairs are coincident objects, colocated chair(1)s and chair(2)s, where a chair(1) cannot survive loss of a leg and a chair(2) can? Or is the question whether I have chair(1)s or chair(2)s settled by figuring out what the majority of the folks in the chair factory thought (suppose my chairs were made in an English speaking country)? Or is it settled by what the bosses thought? Or by what I thought? But I have no opinion on the question. So is there no fact of the matter whether one of the things in my dining room would survive loss of a leg? These questions seem insuperable.
Now one way to get out of all of these puzzles is to be enough of a compositional universalist: any bunch of parts with coherent persistence conditions and interworld identity conditions defines an object.[note 2] Thus, when I come upon the stump, whatever my intentions towards the stump, there is an object there with the persistence conditions of a stump, and an object there with the persistence conditions of a seat. Perhaps, then, I do not bring about the existence of the latter object with my intention, but I simply bring it about that it is appropriate to call the object a "seat". Likewise, while artistic intention is needed to transform the dusty chair into Dusty Chair II, even if I didn't have this intention, there would be a nameless object there with the persistence conditions that Dusty Chair II has.
But the main reason to save the existence of artefacts is to save common sense. And this kind of universalism departs far from common sense. It posits that there is an object sitting in the same chair as me, with the same shape and physical properties as me, but with the counterfactual property that were I to yell "Abracadabra!", it would instantly (faster than light!) move to the Amazon rain forest, where it would be wholly composed of a poison-dart frog. To embrace this kind of compositional universalism to save artefacts seems too costly.
Of course, this denial of the existence of artefacts needs come along with some kind of a paraphrase story that allows ordinary sentences like "She sat on a chair" to be at least approximately true. Presumably, the story will involve particles or fields having chair-wise arrangements, and so on. Moreover, I think we are going to have give such a story even if we admit tables and chairs into our ontology—just not for tables and chairs.
For there are always going to be artefact-like cases where almost nobody—not even the compositional universalist—will want to posit an object. The person whose home is a museum can make a work of art "composed" of shadows, by simply placing a tag on a blank wall with an interesting shadow pattern. But shadows don't exist. Or one could have an earring that is made entirely of the concentration in some kind of a field (maybe a magnetic one—the earring then could be seen by aliens who have a magnetic sense). But while fields might exist, it seems unlikely that concentrations of them do. Just as one can use a nail, it seems one can use a hole (one can put something in it, or one can sell it to an art gallery), but holes don't seem to exist.
Authors write books, and programmers write software. And it seems that, more and more, the most valuable artefacts are of this sort, artefacts that are the subject matter of intellectual property law rather than of tangible property law. These books and pieces of software seem type-like, abstract, rather than object-like. It is the book-type that we care about the author's writing. But what is odd is that if we take these type-like creations seriously, we have to depart from the common-sensical idea that the author causes them to exist. For surely we don't cause type-like things, abstracta, to come into existence.
But suppose we insist that the book the author writes is the manuscript. Where is this manuscript, these days? On a CD sent to the publisher, let us say. The CD is not the manuscript, though. There might be several manuscripts on a CD. The manuscript, it then seems, is made of colored pieces of dye on the CD (assuming a consumer CD-R). Maybe that can be counted an object. But suppose that instead of using a CD, we use some medium where all the data is encoded as the state of some field. Unless we reify states of fields, the data won't be an entity. Or, more simply, let suppose that I put several manuscripts on a CD, and then archive, compress and encrypt them in such a way that it is impossible to single out the bits of one manuscript from the bits of another, but it is possible to decrypt and decompress the archive, and extract the manuscripts. (For instance, the compression may ensure that words and phrases that appear in multiple manuscripts get listed only once in full. The encryption might end up shuffling the bits of all the manuscripts together before applying the encryption function.) Then perhaps there is no plausible way to identify the manuscript with a bunch of colored patches on the CD, but the manuscript is still "there", on the CD.
In other words, unless we are going to have a really bloated ontology, we are anyway going to end up with artefacts that we don't admit in our ontology, artefacts such that we will need some kind of a paraphrase for sentences that treat them as if they were substances. So the kind of story the person who denies the existence of artefacts will need to give about tables and chairs is one that needs to be given anyway about shadows, holes, books and manuscripts. The cost of the story is, therefore, low.