Six weeks ago, most of the people in my orbit couldn't have told you what a large language model was. Now it's the only thing anyone in this industry wants to talk about. Here's my first real read on what ChatGPT changes for creative work, and what it doesn't.
Six weeks that moved faster than the industry
OpenAI put ChatGPT online on November 30, 2022 (OpenAI). No keynote, no ad campaign, just a research preview dropped on a Wednesday. Five days later, Sam Altman tweeted that it had crossed one million users (Altman). I remember reading that number and doing the math on how long it took other tools I’d used to get there. Most never did.
By the time this piece goes up, ChatGPT has been the subject of more group chats, panic-tweets, and agency all-hands than any tool I’ve watched land in my career. Some of that is warranted. A lot of it isn’t. I want to write down where I actually stand, in real time, before the takes calcify into something less honest.
I’ve spent nights this past month feeding it briefs, tone-of-voice prompts, headline options, the kind of work that used to eat an afternoon. Some of it was genuinely useful. A lot of it was serviceable in the way a rough first draft is serviceable, meaning it needed a person to make it actually good. That gap, between "serviceable" and "good," is where I think this whole conversation actually lives.
What I’ve watched the industry get wrong already
The first reaction from a lot of writers and creatives was to roll their eyes. TechCrunch put it plainly in its year-end wrap-up: "Writers everywhere rolled their eyes at the new technology, much like artists did with OpenAI’s DALL-E model" (Wiggers et al.). I understand the instinct. Every few years something shows up promising to replace craft with a shortcut, and every few years the craft outlasts the shortcut.
But I think dismissing this one outright is a mistake, and so is the opposite instinct I’m seeing from some corners of the industry, the ones already writing thinkpieces about the death of the copywriter. Both reactions skip the actual work of figuring out what the tool is good at and what it isn’t. Neither one is defiant. Panic and hype are both just different flavors of not thinking clearly.
“Writers everywhere rolled their eyes at the new technology, much like artists did with OpenAI’s DALL”
E model." — TechCrunch, "ChatGPT: Everything Released From the AI-Powered Chatbot in 2022," Dec. 31, 2022
That eye-roll is worth sitting with for a second. It’s not wrong that the first draft out of ChatGPT rarely sounds like it belongs to anyone. It’s flat. It hedges. It reaches for the safest phrasing in the room, every time, because that’s what it’s built to do. It optimizes for plausible, not for true, and definitely not for bold. If your standard for creative work is "plausible," this tool is a real threat to you. If your standard is higher than that, which ours is, it’s a different conversation entirely.
This isn’t new, it’s just visible now
Here’s the part I think gets lost in the noise: organizations weren’t waiting for ChatGPT to start using AI. McKinsey’s year-end research found that fifty percent of organizations reported using AI in at least one business function in 2022, up from twenty percent back in 2017 (McKinsey & Company). AI has been quietly doing work inside marketing stacks, ad platforms, and production pipelines for years. What changed on November 30 wasn’t the existence of the technology. It was that, for the first time, anyone could open a browser and talk to it directly. That’s what made it a story instead of a spec sheet.
That distinction matters for how we think about it. This isn’t a new front opening up. It’s an old front becoming visible to everyone at once, including a lot of people in creative roles who are encountering AI directly for the first time and understandably asking what it means for their seat at the table.
Where HAUS stands, plainly
I’m not going to pretend I have this fully figured out six weeks in. Nobody does, and anyone telling you otherwise is selling something. But I know what I believe right now, and I want it on the record.
AI sits at our table. It doesn’t run the haus. We’ll use it the way we use any tool that makes the work faster to get to, not the way you use a tool that makes the work for you. A tool drafts. A craftsperson decides. Those are not the same job, and we’re not going to blur the line between them just because the draft got easier to produce.
“AI sits at our table. It doesn’t run the haus.”
David Keyes, Founder & CEO, HAUS XXIV
The people who built this haus were done blending in long before this tool existed. We didn’t build a standard around chasing whatever’s newest. We built it around raising the bar on the work itself, and a faster first draft doesn’t raise anything by itself. What raises it is what a person does with that draft afterward, the judgment, the taste, the willingness to throw the safe version out and build the dangerous one instead.
So here’s where I land, for now, in writing, dated. We’re not panicking, and we’re not pretending this changes nothing. We’re watching closely, testing honestly, and holding the same line we’ve held since day one: the craft is the standard, the tool is not the craftsperson, and the people doing the deciding are still, and will remain, us.
The conversation isn’t finished
I expect to be wrong about parts of this by the time I write about it again. That’s fine. Writing down a clear-eyed first read, in the middle of the noise, is more useful than waiting for certainty that isn’t coming anytime soon. What I won’t do is let the industry’s mood swing from dismissal to panic decide our position for us. We’ll keep building the way we always have, with better tools in the room when they earn it.
