Advocacy Requires an Audience

WHY DOES ADVOCACY REQUIRE AN AUDIENCE?

Advocacy requires an audience because persuasion does not happen inside a document, it happens inside another person's mind, when they read what you wrote, hold it against what they already believe, and decide to update that belief. It does not matter how well-reasoned, well-supported, or well-organized your argument is if nothing you wrote was actually read by someone capable of being moved by it. An argument with no audience persuades no one.

We have written about the shared understanding that persuasion depends on in another article, on the assumption that the gap between you and your audience was a gap in a real person's understanding of your position. An uncounted number of files over the past year have made it clear that this assumption can no longer be taken for granted.

WHAT HAPPENS WHEN A CHATBOT BECOMES THE AUDIENCE INSTEAD OF A PERSON?

Increasingly, correspondence is not being read at all. It is being pasted into a generative AI tool with an instruction along the lines of "say why this is wrong," and whatever comes back is sent as though it were a considered response. That is not a response. It is a rebuttal manufactured to order, and it would have been generated whether the underlying letter was right, wrong, or somewhere in between.

Imagine a Council writes to an Owner explaining, in careful detail, why a proposed alteration cannot proceed without further information about drainage. The Owner does not read past the first paragraph. Instead, the letter is uploaded to a chatbot with the instruction to argue against it, and the output is sent back as the Owner's reply. The reply sounds confident and organized. It also does not engage with the drainage concern at all, because the person on the other end never saw it as a concern to engage with. They saw it as text to defeat.

WHY DOES THIS MAKE A DISPUTE HARDER TO RESOLVE, NOT EASIER?

This creates two distinct problems, and they compound each other. The first is that a chatbot prompted to argue against something will invent whatever it needs to make that argument sound complete, including authority that does not exist. We have now had several files where correspondence confidently cited a decision, by name and citation, as settled authority for a position. When we looked the decision up, it did not exist. Not that it said something different than claimed. It was not on CanLII, not on any other database, at all. It had been generated because it sounded like the kind of case that should exist, not because it did.

The same pattern shows up with bylaws. A Strata will assert that an Owner's conduct contravenes a specific bylaw. The Owner disputes that the bylaw applies, or that it means what the Strata says it means. When we compare the quoted wording against what is actually on file for that Strata, the bylaw is not there. Our understanding is that a Strata's bylaws are the Standard Bylaws under the Strata Property Act, except to the extent different bylaws have actually been filed in the land title office (s.120), and that even an approved amendment has no effect until that filing is done (s.128(2)). A bylaw that reads smoothly, uses the right vocabulary, and sounds exactly like the kind of provision a strata corporation would have, but was never filed for this Strata, is not a bylaw of this Strata at all. It is usually something else entirely: a plausible-sounding provision pulled from somewhere else on the internet, offered up because it fit the shape of what the prompt was asking for.

WHY IS IT EASIER TO GENERATE AN ARGUMENT THAN TO UNDERSTAND ONE?

It used to cost something to respond to correspondence. You had to read it, understand what was actually being claimed, and work out where you agreed and where you did not, before you could write anything back. That cost was doing real work: it forced at least a minimal amount of engagement with the other side's position before you could disagree with it.

That cost has now dropped close to zero. Generating a confident rebuttal takes less effort than reading the letter it responds to, and a rebuttal produced this way looks the same on the page as one a person actually reasoned through. There is no longer any friction pushing either side toward understanding the other, and the gaps of vocabulary, expectation, and responsibility we have written about elsewhere do not just persist, they widen, because neither side is doing the work that used to close them.

HOW SHOULD YOU RESPOND WHEN THE OTHER SIDE'S "ARGUMENT" MIGHT NOT BE THEIRS?

It is important to verify before you respond to what looks like a considered position. Confirm that a cited case actually exists and says what it is claimed to say, rather than assuming a citation that looks correctly formatted is correct. Confirm that a cited bylaw is actually filed against the Strata in question, rather than assuming quoted language is real because it reads like it belongs. Our recommendation is to treat an unverified citation, whether to a case or a bylaw, as a fact to be checked, not as a starting point to be argued with.

It is also worth resisting the urge to answer a generated rebuttal point by point. A document produced to defeat your letter, without anyone having read it first, does not deserve the same engagement as a genuine disagreement, and responding to every line only rewards the practice of generating one. Focus your reply on the facts and the documents that actually govern the dispute, and put the burden back on the other side to engage with those, rather than chasing whatever a chatbot invented next.

There is always the chance that a person will read your response before uploading it to a chatbot. That is why it is important that a response is structured in a way that will grab a readers attention, that uses narrative to persuade, and that is generally persuasive.

AN ARGUMENT WITHOUT AN AUDIENCE PERSUADES NO ONE

None of this means generative AI tools have no place in a dispute. It means that persuasion still depends on something a chatbot cannot supply on either end: a person who actually reads what is in front of them, checks whether it is true, and is willing to be moved if it is. A rebuttal generated in seconds can look exactly like a considered response, but it has not engaged with anything, and it cannot be persuaded of anything either.

A robust and resilient Strata Community resolves its disputes through people who read each other's positions, not through documents traded past each other at machine speed. The tools may keep improving. The requirement that someone on the other end is actually paying attention will not change.