Today is my birthday, and I have a bad cold. The cold is penance for traveling, socializing, dancing myself clean, and overfilling my cup last weekend out east. I am officially entering my mid-40s, the middest years of midlife. Why yes, I am ecstatic about it! No, that's not sarcastic. Anyway, thank you for the kind birthday note; could you hold that card a little farther away from my face so I can read what it says?
Yes, I am aware that I am overdue for a visit to the doctor, but look, I am presented with new content about perimenopause pretty much dailyy, so other than this cold, I'm ready for whatever my age and my body have to throw at me. Evidently I have one more year before I will be forced to watch my life fall off a cliff, and I will become a completely unrecognizable person. It's dramatic, but I accept the challenge. From here on out I'm fully prepared to do the complete ship of Theseus compared with the woman I was before the pandemic and everything else.
I will get back to my series on how LLMs process information, perhaps next week, perhaps next year. In the meantime, let's indulge in some year-end dice-shaking and note-taking.
Know when to fold 'em, know when to run:
Five year-end observations and predictions for 2026
Google loses, Google wins
For the first time in 25 years, I stopped using Google as my default search engine for most tasks. For work-related and general info-seeking tasks, I start with Claude search mode, which bypasses all the content marketing "behavioral research" blather and surfaces the best data for my use case far faster than Google ever did. AI-generated summaries are failures 90% of the time, but when you stay away from easily spammed consumer keyword/prompt modifiers ("best," "top," etc.) and check your sources, gen AI search is sleek and impressive.
Claude is my go-to AI tool across the board, with its stupid sphincter logo. I'm a sucker for the Claude UX and writing style, and it's a decent line-editor and fact-checker. I use it for software support and for prototyping different methods for organizing systems. I don't let it anywhere near my client deliverables, but also, ever so slightly, it's changed how I work.
Of course I have heard how great Gemini is, but I am not in a huge rush to embrace more screen time during the social season. Software exploration is for Aquarius season, not Sagittarius season. Google will always be there, like CBS, 3M, and AT&T.
My 2026 prediction: Google will release a new product, or seventeen new products. Some people will use them immediately, and some people will not because they do not like Google or otherwise. But whatever Google does, it won't take the damn "analyze my data" AI button out of Google sheets, even though the function has never once worked correctly.
Cultivating the style differentiator
My first assignment in my first semester of grad school was to outline how to measure a concept that was not immediately quantifiable, and since I'd spent the previous year professionally copy-editing, I chose "style." In analog culture, a defined style has long been the primary differentiation in creative industries, and I'm glad to see the central concepts of style measurement emerging in AI-related discourse.
Style is the manifest expression of taste, but we all know people who have excellent taste and aren't stylish. In publishing and in enterprise content strategy, style is the system of applying taste and consistency to output. Style comprises voice, tone, cadence, messaging strategy, and information architecture. Style guides define the clichés we embrace and the constructions we abhor. And, as most of the readers of this newsletter know, in both the business and publishing worlds, style guides are undervalued and under-enforced.
Style protects writers, too! Last month Politico was slapped on the wrist because it implemented AI tools that did not follow the house style guide. Cheers to NewsGuild-CWA for making a documented style guide court-admissible evidence against sloppy AI! And it's not that generative AI can't use style guides (it can and should!), but AI systems don't magically pick up on house style without proper implementation.
And compared with print, it's long been hard to find good style on the web, where usability, device compatibility, and the constraints of code force websites and software to look blocky. In the 2010s, publications invested in artificially generated traffic and optimization of on-page advertising instead of remembering why audiences connect with publications to begin with. One of the (few) positive effects of the creator economy is that publishers are rediscovering (I think?) how a niche or "authentic" perspective beats chasing consensus trends and penny-pincher traffic.
And after three years of claiming ChatGPT will soon replace writers, I'm glad popular culture is reckoning with generative AI style. I'd also like to add that most human internet writing is mediocre at the sentence level, and yes, I am talking about your publication and your posts. Mine, too, sure, but I'm not asking for payment, and you might be. We could all take a hard look at how we can improve our style.
My 2026 prediction: Major publishers and brands will reinvest in style, usability, and trustworthiness as a growth driver.
Kicking the can down the road in re: IP
I'm not a lawyer, so take my advice with all the grains of salt, but I do not believe contemporary intellectual property law is going to save anyone's independent publishing business from AI copyright infringement, especially not in the current political environment. If the NYT vs. OpenAI copyright case is litigated this year, maybe we will see a few more answers, but I am not holding out hope for intellectual property law resurgence among all the media mega-mergers, especially when Disney bent over so very quickly.
If you are an independent creator or a small publisher, and you want to publish text on the internet, assume someone will take your intellectual property and clone it or use it to build a dataset to train AI. It doesn't matter if you have paid content; someone will still pay the money, scrape the content, and put it in their dataset. They do not believe that your words are valuable, or that you worked really hard. They do not think your ideas are earth-shattering, and they definitely don't want to pay you. The internet has always worked this way, but it's far easier to scale and spam with this approach now. Why did this happen? Blame everyone who trusted that Creative Commons and robots.txt files would protect their IP back in 2008. If the problem's not solved by this point, it's not going to be.
But AI systems for content discovery are here to stay. Google has been working toward this outcome (and being clear about this outcome) for years. LLMs will be a gateway to discovery of your brand, if not the primary gateway, and if you're going to get weird about charging for crawls, most AI systems will ignore your voice... and that's neither good for your business or the information ecosystem more broadly. Information quality is bad in AI, if you haven't noticed, and as social feed inventories shift toward misinformation, good publishers have a chance to stand out, regardless of whether AI has access to their IP.
Generative AI is not popularly regarded as trustworthy. Yes, some people will be fine with consuming sloppy stolen and repurposed content. But plenty of others are vocally seeking alternatives, and I hope more publishers work harder to reach them and meet their needs, rather than following the IP red herring.
My 2026 prediction: A few publishers will adjust their business models to compete with AI experiences. Most will spiral out for a while until AI-native publications grow. No one will win much money from IP, ever.
Knowledge graphs: Still on my high horse
I haven't yet written the knowledge graph part of my AI explainer essay series, which is a shame, because knowledge graphs are my favorite part of the whole ecosystem. Knowledge graphs are mapped semantic sources of truth that AI-based systems can access, and I am betting on them for the near future.
My 2026 prediction: I will learn even more about knowledge graphs, and I'll pass some of that on.
Strengthen search and build your own recommender system
If you are an SEO or editor at a company that uses an AI-enabled tool like Algolia or Shaped for on-site search, find a way to work on or work with the internal search product. You will learn more about how AI-powered search systems work than most folks who have been working in SEO for years.
My 2026 prediction: At least one major publisher will build their own machine learning-enabled content feed designed to compete with mass social media. At least one startup publisher feed powered by the same technology will compete with legacy media.
Scenes, not communities
Due to my anti-cult stance, I will not join any community fronted by a single, charismatic leader, and that includes every solo internet creator enterprise. I am fearful of the dynamics of group chats and book clubs. The hustle bits of the Creator Economy are immensely tiring. But I would love to happen upon more scenes, or loosely organized groups of writers, artists, and creators who support and recommend each other in a respectful, collegial way without pre-existing social media promotional expectations or having to call something "core" or to define the vibes. I like a scene.
My 2026 prediction: I will discover the cool scene you're a part of, because you'll reply to this email and tell me about it.
Welp. I've fizzled into a sinus headache. Now that I've pretended to be a real writer for a few hours, I will end my birthday with pizza, the final episode of The Chair Company, and listening to records with the hubs and the cats. I will take stock of where I am and get ready to fight another year. My casual goal was to get to newsletter 250 this year... and I can probably conjure that before I abscond to the tropics next week.
Cheers,
DC
The Content Technologist is a newsletter and consultancy based in Minneapolis, working with clients and collaborators around the world. The entire newsletter is written and edited by Deborah Carver, an independent content strategy consultant.
Affiliate referrals: Ghost publishing system | Bonsai contract/invoicing | The Sample newsletter exchange referral | Writer AI Writing Assistant
Cultural recommendations / personal social: Spotify | Instagram | Letterboxd | PI.FYI
Did you read? is the assorted content at the very bottom of the email. Cultural recommendations, off-kilter thoughts, and quotes from foundational works of media theory we first read in college—all fair game for this section.
My favorite songs of 2025 are listed here. Yes, I know Spotify sucks—currently reading Liz Pelly's excellently researched and written Mood Machine—and I'm going to migrate to an alternative, but that's also an Aquarius season task.
