You are correct that the a.c.i. (and some other constructions) is called indirect speech. So Lucretius could have written, in a new sentence, nequaquam divinitus est creata natura, which would not have been indirect speech. But I think the main point here is that indirect speech is not necessarily speech from someone other than the speaker: it can be the same person.
The function of indirect speech can be to express the attitude of the speaker towards a certain statement or phrase. In this case, one could say that Lucretius expresses epistemic modality towards his theory and tells us how he come to it: he demonstrates or asserts (confirmare) and recounts (reddo) the content of his theory, the indirect speech. He takes position with regard to the theory.
Had he phrased it in direct speech, it would have been presented as a fact by default, which was not his intention. He could have qualified indirect speech in other ways to get around that; but indirect speech is a natural approach.
Compare this with how one would write the same thing in English:
I assert that the gods were not involved in the creation of nature.
This is indirect speech as well, marked by the introductory word that. It's the natural way if you want to use a word like assert.
On a side note, the term 'direct speech' means speech exactly as it is phrased by the speaker (or m. m. the writer). In a way, the entire De Rerum Natura, like any text, is direct speech, as one can quote from it; it contains words written exactly so by Lucretius. Then indirect speech is embedded in the direct speech.