The English language is only defined by its usage. If you feel that is a matter of opinion, then it is an opinion that is shared by the major authoritative references on the language. OED, Cambridge, Merriam-Webster, etc. all seek to only document usage, rather than dictate definitions.
I think it’s terribly ironic that this argument revolves around robot vs bot and the applicability to software. Robot is a relatively new word (Google says 1920s), and software is maybe even a newer concept (in the broader public psyche, if we discount 19th century punch card looms/hypothetical difference engines as being too niche). The shortening of robot to “bot” is well established and the perceived rule that only “bot” can be applied to software could only from an observed bias from very recent usage (and even the universality of that is suspect, e.g., see the “robots.txt” example that I mention above).
Basically, you are arguing for an apparent “rule” that most likely only emerged from very recent usage (in the form of a common, but not universal bias).
Popular usage can truly effect change, and accepting this can help avoid experiencing a negative affect.