What does GenAI mean for product marketing?

Since one of the core deliverable of product marketing is content (i.e., what does owning product messaging mean if you don't have content to implement it into GTM?), I wondered if generative AI would finally come for my job. After playing around with ChatGPT a bit and seeing how others are using it, I am now ready to render the definitive judgement: generative AI will accelerate content extensions/derivations off an existing piece of content, but the primary source content will still rely on skillful product marketers.

For example, if you have a prewritten blog post, GenAI will be great for writing a summary blurb or webinar abstract, and probably even creating display banners to promote the content assuming you have well-formed design guidelines, but at least for the next 2-3 years, PMMs should approach GenAI as an accelerant to creating derivative work rather than as a real job threat for two big reasons:

1. GenAI can be believably wrong: LLMs don't actually "know" anything. Rather they pattern match based on the training data. What makes factual mistakes coming from an LLM had to identify is precisely because they seem believable and therefore a mistake is easier to miss. LLMs have been known to make up citations that look real. Or else a model fully "trusts" content it's trained on with no regards to credibility (interesting read here about how LLMs are about twenty years behind compared to search engines when it comes to analyzing not just content, but also "authority"). There's no faster way to lose credibility with a buyer than getting a basic fact wrong - e.g., a GenAI piece incorrectly extrapolates what protocols or frameworks your product supports.
2. You're selling to knowledgeable buyers: The first reason is probably fixed over time as data scientists learn more about how GenAI is being used and how black hat actors try to exploit it. But for me the largest impediment is more conceptual: If PMM was just about summarizing product features, then AI could easily replace that skill now. But in tech, your buyer is usually a technical decision maker who has spent decades in their field. Good content will teach them something new or make them think of a problem differently. Pattern matching based on what has already been created before is precisely what buyers ignore.