Generative AI (or GenAI for the cool kids) has shaken up a lot of industries. Because large-language models (LLMs) deal primarily with language as their first love, it’s no surprise that it feels poised to replace human writers in particular.
Now, a calm breath will remind us that this new technology is very, well, new, and it will evolve rapidly as the current arms race screams forward. So maybe it won’t replace all writers tomorrow. But even if that’s true, what about in, say, five years?
First of all, no one really knows. Few saw this coming, and we’re all still sorting through what it means. The interwebs are full of speculation, and whatever your opinion is, you can probably find someone out there to back you up.
But it’s hard to imagine that everything will remain the same in the face of this most disruptive of technologies. For example, I’ve seen some of the summaries that AI produces, and as much as I’d like to sneer at it, it can actually do a pretty good job. I do interviews and look at the automated summaries, and I’m pretty impressed.
So there will definitely be things that AI will do well. What are the frontiers, then, that will be more challenging? Well, it’s important to remember that these models are trained on vast reams of existing information plus statistics on what words are likely to go together. You might say this makes LLMs really good at reassembling or restating things that have already been said. Yes, there’s talk of “emergent” properties that aren’t well understood that allow the algorithms to venture outside of their bounds. But for the most part, it’s an engine that processes things that some humans somewhere already know.
If that’s what you require, AI may indeed do a good job for you — if not today, then down the road. But most companies in our industry are in the business of doing things that haven’t been done before. At risk of using a severely overused word, that’s what innovation is all about. Yes, you can train models on your technology, but that’s a far narrower corpus than the entirety of the web, so it will be harder for it to bring together the necessary technical, business, management, and marketing considerations.
That’s what humans can do best. Those initial press articles and product pitches and presentations and whitepapers you use to communicate your ideas to your customers are best done by a writer with good technical, management, and marketing underpinnings who isn’t too close to the technology. Such a writer can be an interface between your internal knowledge and the external world.
Written narrative pieces that flow and communicate your technology from several angles are still a human endeavor and will likely remain so for quite a while. If you’re debating the best way to approach your public-facing documentation, give me a shout. The rumors of human writers’ demise are, so far, greatly exaggerated.